Export grenad operations to file

4400: Upgrade rustls to 0.21.10 and ring to 0.17 r=curquiza a=hack3ric

# Pull Request

## What does this PR do?
- Upgrade dependencies that uses ring 0.16 so that they rely on ring 0.17 instead
- Use rustls 0.21 for actix-{http,tls}, since newer versions of rustls uses ring 0.17
- Fix some trivial breaking API changes caused by above

## PR checklist
Please check if your PR fulfills the following requirements:
- [x] Does this PR fix an existing issue, or have you listed the changes applied in the PR description (and why they are needed)?
- [x] Have you read the contributing guidelines?
- [x] Have you made sure that the title is accurate and descriptive of the changes?

Thank you so much for contributing to Meilisearch!


Co-authored-by: Eric Long <i@hack3r.moe>

.cargo/config.toml

@@ -0,0 +1,2 @@
+[alias]
+xtask = "run --package xtask --"

.github/ISSUE_TEMPLATE/sprint_issue.md

@@ -27,6 +27,23 @@ Related spec: WIP
 - [ ] If prototype validated, merge changes into `main`
 - [ ] Update the spec
 
+### Reminders when modifying the Setting API
+
+<!--- Special steps to remind when adding a new index setting -->
+
+- [ ] Ensure the new setting route is at least tested by the [`test_setting_routes` macro](https://github.com/meilisearch/meilisearch/blob/5204c0b60b384cbc79621b6b2176fca086069e8e/meilisearch/tests/settings/get_settings.rs#L276)
+- [ ] Ensure Analytics are fully implemented
+  - [ ] `/settings/my-new-setting` configurated in the [`make_setting_routes` macro](https://github.com/meilisearch/meilisearch/blob/5204c0b60b384cbc79621b6b2176fca086069e8e/meilisearch/src/routes/indexes/settings.rs#L141-L165)
+  - [ ] global `/settings` route configurated in the [`update_all` function](https://github.com/meilisearch/meilisearch/blob/5204c0b60b384cbc79621b6b2176fca086069e8e/meilisearch/src/routes/indexes/settings.rs#L655-L751)
+- [ ] Ensure the dump serializing is consistent with the `/settings` route serializing, e.g., enums case can be different (`camelCase` in route and `PascalCase` in the dump)
+
+#### Special cases when adding a setting for an experimental feature
+
+- [ ] ⚠️ API stability: The setting does not appear on the main settings route when the feature has never been enabled (e.g. mark it `Unset` when returned from the index in this situation. See [an example](https://github.com/meilisearch/meilisearch/blob/7a89abd2a025606a42f8b219e539117eb2eb029f/meilisearch-types/src/settings.rs#L608))
+- [ ] The setting cannot be set when the feature is disabled, either by the main settings route or the subroute (see [`validate_settings` function](https://github.com/meilisearch/meilisearch/blob/7a89abd2a025606a42f8b219e539117eb2eb029f/meilisearch/src/routes/indexes/settings.rs#L811))
+- [ ] If possible, the setting is reset when the feature is disabled (hard if it requires reindexing)
+
 ## Impacted teams
 
 <!---Ping the related teams. Ask for the engine manager if any hesitation-->
+<!---@meilisearch/docs-team when there is any API change, e.g. settings addition-->

.github/workflows/publish-docker-images.yml

@@ -97,7 +97,7 @@ jobs:
       - name: Send CI information to Cloud team
         # Do not send if nightly build (i.e. 'schedule' or 'workflow_dispatch' event)
         if: github.event_name == 'push'
-        uses: peter-evans/repository-dispatch@v2
+        uses: peter-evans/repository-dispatch@v3
         with:
           token: ${{ secrets.MEILI_BOT_GH_PAT }}
           repository: meilisearch/meilisearch-cloud

.github/workflows/sdks-tests.yml

@@ -22,7 +22,7 @@ jobs:
     outputs:
       docker-image: ${{ steps.define-image.outputs.docker-image }}
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - name: Define the Docker image we need to use
         id: define-image
         run: |
@@ -46,11 +46,11 @@ jobs:
       MEILISEARCH_VERSION: ${{ needs.define-docker-image.outputs.docker-image }}
 
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
         with:
           repository: meilisearch/meilisearch-dotnet
       - name: Setup .NET Core
-        uses: actions/setup-dotnet@v3
+        uses: actions/setup-dotnet@v4
         with:
           dotnet-version: "6.0.x"
       - name: Install dependencies
@@ -75,12 +75,12 @@ jobs:
         ports:
           - '7700:7700'
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
         with:
           repository: meilisearch/meilisearch-dart
       - uses: dart-lang/setup-dart@v1
         with:
-          sdk: 3.1.1
+          sdk: 'latest'
       - name: Install dependencies
         run: dart pub get
       - name: Run integration tests
@@ -100,10 +100,10 @@ jobs:
           - '7700:7700'
     steps:
       - name: Set up Go
-        uses: actions/setup-go@v4
+        uses: actions/setup-go@v5
         with:
           go-version: stable
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
         with:
           repository: meilisearch/meilisearch-go
       - name: Get dependencies
@@ -129,11 +129,11 @@ jobs:
         ports:
           - '7700:7700'
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
         with:
           repository: meilisearch/meilisearch-java
       - name: Set up Java
-        uses: actions/setup-java@v3
+        uses: actions/setup-java@v4
         with:
           java-version: 8
           distribution: 'zulu'
@@ -156,7 +156,7 @@ jobs:
         ports:
           - '7700:7700'
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
         with:
           repository: meilisearch/meilisearch-js
       - name: Setup node
@@ -191,7 +191,7 @@ jobs:
         ports:
           - '7700:7700'
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
         with:
           repository: meilisearch/meilisearch-php
       - name: Install PHP
@@ -220,11 +220,11 @@ jobs:
         ports:
           - '7700:7700'
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
         with:
           repository: meilisearch/meilisearch-python
       - name: Set up Python
-        uses: actions/setup-python@v4
+        uses: actions/setup-python@v5
       - name: Install pipenv
         uses: dschep/install-pipenv-action@v1
       - name: Install dependencies
@@ -245,7 +245,7 @@ jobs:
         ports:
           - '7700:7700'
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
         with:
           repository: meilisearch/meilisearch-ruby
       - name: Set up Ruby 3
@@ -270,7 +270,7 @@ jobs:
         ports:
           - '7700:7700'
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
         with:
           repository: meilisearch/meilisearch-rust
       - name: Build
@@ -291,7 +291,7 @@ jobs:
         ports:
           - '7700:7700'
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
         with:
           repository: meilisearch/meilisearch-swift
       - name: Run tests
@@ -314,7 +314,7 @@ jobs:
         ports:
           - '7700:7700'
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
         with:
           repository: meilisearch/meilisearch-js-plugins
       - name: Setup node
@@ -345,7 +345,7 @@ jobs:
         ports:
           - '7700:7700'
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
         with:
           repository: meilisearch/meilisearch-rails
       - name: Set up Ruby 3
@@ -369,7 +369,7 @@ jobs:
         ports:
           - '7700:7700'
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
         with:
           repository: meilisearch/meilisearch-symfony
       - name: Install PHP

.github/workflows/test-suite.yml

@@ -66,6 +66,10 @@ jobs:
       - uses: actions/checkout@v3
       - name: Cache dependencies
         uses: Swatinem/rust-cache@v2.7.1
+      - uses: actions-rs/toolchain@v1
+        with:
+          toolchain: stable
+          override: true
       - name: Run cargo check without any default features
         uses: actions-rs/cargo@v1
         with:
@@ -78,7 +82,7 @@ jobs:
           args: --locked --release --all
 
   test-all-features:
-    name: Tests all features
+    name: Tests almost all features
     runs-on: ubuntu-latest
     container:
       # Use ubuntu-18.04 to compile with glibc 2.27, which are the production expectations
@@ -94,16 +98,12 @@ jobs:
         with:
           toolchain: stable
           override: true
-      - name: Run cargo build with all features
-        uses: actions-rs/cargo@v1
-        with:
-          command: build
-          args: --workspace --locked --release --all-features
-      - name: Run cargo test with all features
-        uses: actions-rs/cargo@v1
-        with:
-          command: test
-          args: --workspace --locked --release --all-features
+      - name: Run cargo build with almost all features
+        run: |
+          cargo build --workspace --locked --release --features "$(cargo xtask list-features --exclude-feature cuda)"
+      - name: Run cargo test with almost all features
+        run: |
+          cargo test --workspace --locked --release --features "$(cargo xtask list-features --exclude-feature cuda)"
 
   test-disabled-tokenization:
     name: Test disabled tokenization
@@ -164,7 +164,7 @@ jobs:
       - uses: actions-rs/toolchain@v1
         with:
           profile: minimal
-          toolchain: 1.71.1
+          toolchain: 1.75.0
           override: true
           components: clippy
       - name: Cache dependencies

CONTRIBUTING.md

@@ -75,6 +75,12 @@ If you get a "Too many open files" error you might want to increase the open fil
 ulimit -Sn 3000
 ```
 
+#### Build tools
+
+Meilisearch follows the [cargo xtask](https://github.com/matklad/cargo-xtask) workflow to provide some build tools.
+
+Run `cargo xtask --help` from the root of the repository to find out what is available.
+
 ## Git Guidelines
 
 ### Git Branches

Cargo.toml

@@ -16,11 +16,16 @@ members = [
     "json-depth-checker",
     "benchmarks",
     "fuzzers",
+    "tracing-trace",
+    "xtask",
 ]
 
 [workspace.package]
-version = "1.5.0"
-authors = ["Quentin de Quelen <quentin@dequelen.me>", "Clément Renault <clement@meilisearch.com>"]
+version = "1.7.0"
+authors = [
+    "Quentin de Quelen <quentin@dequelen.me>",
+    "Clément Renault <clement@meilisearch.com>",
+]
 description = "Meilisearch HTTP server"
 homepage = "https://meilisearch.com"
 readme = "README.md"

Dockerfile

@@ -1,5 +1,5 @@
 # Compile
-FROM    rust:alpine3.16 AS compiler
+FROM    rust:1.75.0-alpine3.18 AS compiler
 
 RUN     apk add -q --update-cache --no-cache build-base openssl-dev
 

LICENSE

@@ -1,6 +1,6 @@
 MIT License
 
-Copyright (c) 2019-2022 Meili SAS
+Copyright (c) 2019-2024 Meili SAS
 
 Permission is hereby granted, free of charge, to any person obtaining a copy
 of this software and associated documentation files (the "Software"), to deal

README.md

@@ -41,10 +41,10 @@ Meilisearch helps you shape a delightful search experience in a snap, offering f
 ## ✨ Features
 
 - **Search-as-you-type:** find search results in less than 50 milliseconds
-- **[Typo tolerance](https://www.meilisearch.com/docs/learn/getting_started/customizing_relevancy?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=features#typo-tolerance):** get relevant matches even when queries contain typos and misspellings
-- **[Filtering](https://www.meilisearch.com/docs/learn/fine_tuning_results/filtering?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=features) and [faceted search](https://www.meilisearch.com/docs/learn/fine_tuning_results/faceted_search?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=features):** enhance your user's search experience with custom filters and build a faceted search interface in a few lines of code
+- **[Typo tolerance](https://www.meilisearch.com/docs/learn/configuration/typo_tolerance?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=features):** get relevant matches even when queries contain typos and misspellings
+- **[Filtering](https://www.meilisearch.com/docs/learn/fine_tuning_results/filtering?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=features) and [faceted search](https://www.meilisearch.com/docs/learn/fine_tuning_results/faceted_search?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=features):** enhance your users' search experience with custom filters and build a faceted search interface in a few lines of code
 - **[Sorting](https://www.meilisearch.com/docs/learn/fine_tuning_results/sorting?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=features):** sort results based on price, date, or pretty much anything else your users need
-- **[Synonym support](https://www.meilisearch.com/docs/learn/getting_started/customizing_relevancy?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=features#synonyms):** configure synonyms to include more relevant content in your search results
+- **[Synonym support](https://www.meilisearch.com/docs/learn/configuration/synonyms?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=features):** configure synonyms to include more relevant content in your search results
 - **[Geosearch](https://www.meilisearch.com/docs/learn/fine_tuning_results/geosearch?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=features):** filter and sort documents based on geographic data
 - **[Extensive language support](https://www.meilisearch.com/docs/learn/what_is_meilisearch/language?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=features):** search datasets in any language, with optimized support for Chinese, Japanese, Hebrew, and languages using the Latin alphabet
 - **[Security management](https://www.meilisearch.com/docs/learn/security/master_api_keys?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=features):** control which users can access what data with API keys that allow fine-grained permissions handling
@@ -61,8 +61,6 @@ You can consult Meilisearch's documentation at [https://www.meilisearch.com/docs
 
 For basic instructions on how to set up Meilisearch, add documents to an index, and search for documents, take a look at our [Quick Start](https://www.meilisearch.com/docs/learn/getting_started/quick_start?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=get-started) guide.
 
-You may also want to check out [Meilisearch 101](https://www.meilisearch.com/docs/learn/getting_started/filtering_and_sorting?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=get-started) for an introduction to some of Meilisearch's most popular features.
-
 ## ⚡ Supercharge your Meilisearch experience
 
 Say goodbye to server deployment and manual updates with [Meilisearch Cloud](https://www.meilisearch.com/cloud?utm_campaign=oss&utm_source=github&utm_medium=meilisearch). No credit card required.
@@ -101,7 +99,7 @@ Meilisearch is a search engine created by [Meili](https://www.welcometothejungle
 
 - For feature requests, please visit our [product repository](https://github.com/meilisearch/product/discussions)
 - Found a bug? Open an [issue](https://github.com/meilisearch/meilisearch/issues)!
-- Want to be part of our Discord community? [Join us!](https://discord.gg/meilisearch)
+- Want to be part of our Discord community? [Join us!](https://discord.meilisearch.com/?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=contact)
 
 Thank you for your support!
 

assets/grafana-dashboard.json

@@ -106,7 +106,7 @@
           },
           "editorMode": "builder",
           "exemplar": true,
-          "expr": "meilisearch_index_count{job=\"meilisearch\", instance=\"$instance\"}",
+          "expr": "meilisearch_index_count{job=\"$job\", instance=\"$instance\"}",
           "interval": "",
           "legendFormat": "",
           "range": true,
@@ -165,7 +165,7 @@
             "type": "prometheus"
           },
           "editorMode": "builder",
-          "expr": "meilisearch_index_docs_count{job=\"meilisearch\", index=\"$Index\", instance=\"$instance\"}",
+          "expr": "meilisearch_index_docs_count{job=\"$job\", index=\"$Index\", instance=\"$instance\"}",
           "hide": false,
           "range": true,
           "refId": "A"
@@ -228,7 +228,7 @@
           },
           "editorMode": "builder",
           "exemplar": true,
-          "expr": "round(increase(meilisearch_http_requests_total{method=\"POST\", path=\"/indexes/$Index/search\", job=\"meilisearch\"}[1h]))",
+          "expr": "round(increase(meilisearch_http_requests_total{method=\"POST\", path=\"/indexes/$Index/search\", job=\"$job\"}[1h]))",
           "interval": "",
           "legendFormat": "",
           "range": true,
@@ -288,7 +288,7 @@
           },
           "editorMode": "builder",
           "exemplar": true,
-          "expr": "round(increase(meilisearch_http_requests_total{method=\"POST\", path=\"/indexes/$Index/search\", job=\"meilisearch\"}[24h]))",
+          "expr": "round(increase(meilisearch_http_requests_total{method=\"POST\", path=\"/indexes/$Index/search\", job=\"$job\"}[24h]))",
           "interval": "",
           "legendFormat": "",
           "range": true,
@@ -348,7 +348,7 @@
           },
           "editorMode": "builder",
           "exemplar": true,
-          "expr": "round(increase(meilisearch_http_requests_total{method=\"POST\", path=\"/indexes/$Index/search\", job=\"meilisearch\"}[30d]))",
+          "expr": "round(increase(meilisearch_http_requests_total{method=\"POST\", path=\"/indexes/$Index/search\", job=\"$job\"}[30d]))",
           "interval": "",
           "legendFormat": "",
           "range": true,
@@ -447,7 +447,7 @@
           },
           "editorMode": "builder",
           "exemplar": true,
-          "expr": "meilisearch_db_size_bytes{job=\"meilisearch\", instance=\"$instance\"}",
+          "expr": "meilisearch_db_size_bytes{job=\"$job\", instance=\"$instance\"}",
           "interval": "",
           "legendFormat": "Database size on disk",
           "range": true,
@@ -458,7 +458,7 @@
             "type": "prometheus"
           },
           "editorMode": "builder",
-          "expr": "meilisearch_used_db_size_bytes{job=\"meilisearch\", instance=\"$instance\"}",
+          "expr": "meilisearch_used_db_size_bytes{job=\"$job\", instance=\"$instance\"}",
           "hide": false,
           "legendFormat": "Used bytes",
           "range": true,
@@ -553,7 +553,7 @@
           },
           "editorMode": "builder",
           "exemplar": true,
-          "expr": "rate(meilisearch_http_response_time_seconds_sum{instance=\"$instance\", job=\"meilisearch\"}[5m]) / rate(meilisearch_http_response_time_seconds_count[5m])",
+          "expr": "rate(meilisearch_http_response_time_seconds_sum{instance=\"$instance\", job=\"$job\"}[5m]) / rate(meilisearch_http_response_time_seconds_count[5m])",
           "interval": "",
           "legendFormat": "{{method}} {{path}}",
           "range": true,
@@ -646,7 +646,7 @@
           },
           "editorMode": "builder",
           "exemplar": true,
-          "expr": "rate(meilisearch_http_requests_total{instance=\"$instance\", job=\"meilisearch\"}[5m])",
+          "expr": "rate(meilisearch_http_requests_total{instance=\"$instance\", job=\"$job\"}[5m])",
           "interval": "",
           "legendFormat": "{{method}} {{path}}",
           "range": true,
@@ -744,7 +744,7 @@
           },
           "editorMode": "builder",
           "exemplar": true,
-          "expr": "sum by(le) (increase(meilisearch_http_response_time_seconds_bucket{path=\"/indexes/$Index/search\", instance=\"$instance\", job=\"meilisearch\"}[30s]))",
+          "expr": "sum by(le) (increase(meilisearch_http_response_time_seconds_bucket{path=\"/indexes/$Index/search\", instance=\"$instance\", job=\"$job\"}[30s]))",
           "format": "heatmap",
           "interval": "",
           "legendFormat": "{{le}}",
@@ -854,7 +854,7 @@
           },
           "editorMode": "builder",
           "exemplar": true,
-          "expr": "meilisearch_nb_tasks{instance=\"$instance\", job=\"meilisearch\", kind=\"statuses\"}",
+          "expr": "meilisearch_nb_tasks{instance=\"$instance\", job=\"$job\", kind=\"statuses\"}",
           "interval": "",
           "legendFormat": "{{value}} ",
           "range": true,
@@ -947,7 +947,7 @@
           },
           "editorMode": "builder",
           "exemplar": true,
-          "expr": "meilisearch_nb_tasks{instance=\"$instance\", job=\"meilisearch\", kind=\"types\"}",
+          "expr": "meilisearch_nb_tasks{instance=\"$instance\", job=\"$job\", kind=\"types\"}",
           "interval": "",
           "legendFormat": "{{value}} ",
           "range": true,
@@ -1040,7 +1040,7 @@
           },
           "editorMode": "builder",
           "exemplar": true,
-          "expr": "meilisearch_nb_tasks{instance=\"$instance\", job=\"meilisearch\", kind=\"indexes\"}",
+          "expr": "meilisearch_nb_tasks{instance=\"$instance\", job=\"$job\", kind=\"indexes\"}",
           "interval": "",
           "legendFormat": "{{value}} ",
           "range": true,
@@ -1161,7 +1161,7 @@
           },
           "editorMode": "builder",
           "exemplar": true,
-          "expr": "rate(process_cpu_seconds_total{job=\"meilisearch\", instance=\"$instance\"}[1m])",
+          "expr": "rate(process_cpu_seconds_total{job=\"$job\", instance=\"$instance\"}[1m])",
           "interval": "",
           "legendFormat": "process",
           "range": true,
@@ -1264,7 +1264,7 @@
           },
           "editorMode": "builder",
           "exemplar": true,
-          "expr": "process_resident_memory_bytes{job=\"meilisearch\", instance=\"$instance\"} / 1024 / 1024",
+          "expr": "process_resident_memory_bytes{job=\"$job\", instance=\"$instance\"} / 1024 / 1024",
           "interval": "",
           "legendFormat": "process",
           "range": true,
@@ -1342,6 +1342,33 @@
         "skipUrlSync": false,
         "sort": 0,
         "type": "query"
+      },
+      {
+        "current": {
+          "selected": true,
+          "text": "meilisearch",
+          "value": "meilisearch"
+        },
+        "datasource": {
+          "type": "prometheus"
+        },
+        "definition": "label_values(job)",
+        "description": "Prometheus job_name from scrape config (default is meilisearch)",
+        "hide": 0,
+        "includeAll": false,
+        "label": "Job",
+        "multi": false,
+        "name": "job",
+        "options": [],
+        "query": {
+          "query": "label_values(job)",
+          "refId": "StandardVariableQuery"
+        },
+        "refresh": 1,
+        "regex": "",
+        "skipUrlSync": false,
+        "sort": 0,
+        "type": "query"
       }
     ]
   },

benchmarks/Cargo.toml

@@ -11,24 +11,24 @@ edition.workspace = true
 license.workspace = true
 
 [dependencies]
-anyhow = "1.0.70"
-csv = "1.2.1"
+anyhow = "1.0.79"
+csv = "1.3.0"
 milli = { path = "../milli" }
-mimalloc = { version = "0.1.37", default-features = false }
-serde_json = { version = "1.0.95", features = ["preserve_order"] }
+mimalloc = { version = "0.1.39", default-features = false }
+serde_json = { version = "1.0.111", features = ["preserve_order"] }
 
 [dev-dependencies]
 criterion = { version = "0.5.1", features = ["html_reports"] }
 rand = "0.8.5"
 rand_chacha = "0.3.1"
-roaring = "0.10.1"
+roaring = "0.10.2"
 
 [build-dependencies]
-anyhow = "1.0.70"
-bytes = "1.4.0"
+anyhow = "1.0.79"
+bytes = "1.5.0"
 convert_case = "0.6.0"
-flate2 = "1.0.25"
-reqwest = { version = "0.11.16", features = ["blocking", "rustls-tls"], default-features = false }
+flate2 = "1.0.28"
+reqwest = { version = "0.11.23", features = ["blocking", "rustls-tls"], default-features = false }
 
 [features]
 default = ["milli/all-tokenizations"]

config.toml

@@ -129,3 +129,6 @@ experimental_enable_metrics = false
 
 # Experimental RAM reduction during indexing, do not use in production, see: <https://github.com/meilisearch/product/discussions/652>
 experimental_reduce_indexing_memory_usage = false
+
+# Experimentally reduces the maximum number of tasks that will be processed at once, see: <https://github.com/orgs/meilisearch/discussions/713>
+# experimental_max_number_of_batched_tasks = 100

dump/Cargo.toml

@@ -11,22 +11,22 @@ readme.workspace = true
 license.workspace = true
 
 [dependencies]
-anyhow = "1.0.70"
-flate2 = "1.0.25"
-http = "0.2.9"
-log = "0.4.17"
+anyhow = "1.0.79"
+flate2 = "1.0.28"
+http = "0.2.11"
 meilisearch-auth = { path = "../meilisearch-auth" }
 meilisearch-types = { path = "../meilisearch-types" }
-once_cell = "1.17.1"
-regex = "1.7.3"
-roaring = { version = "0.10.1", features = ["serde"] }
-serde = { version = "1.0.160", features = ["derive"] }
-serde_json = { version = "1.0.95", features = ["preserve_order"] }
-tar = "0.4.38"
-tempfile = "3.5.0"
-thiserror = "1.0.40"
-time = { version = "0.3.20", features = ["serde-well-known", "formatting", "parsing", "macros"] }
-uuid = { version = "1.3.1", features = ["serde", "v4"] }
+once_cell = "1.19.0"
+regex = "1.10.2"
+roaring = { version = "0.10.2", features = ["serde"] }
+serde = { version = "1.0.195", features = ["derive"] }
+serde_json = { version = "1.0.111", features = ["preserve_order"] }
+tar = "0.4.40"
+tempfile = "3.9.0"
+thiserror = "1.0.56"
+time = { version = "0.3.31", features = ["serde-well-known", "formatting", "parsing", "macros"] }
+tracing = "0.1.40"
+uuid = { version = "1.6.1", features = ["serde", "v4"] }
 
 [dev-dependencies]
 big_s = "1.0.2"

dump/src/lib.rs

@@ -276,6 +276,7 @@ pub(crate) mod test {
                 ),
             }),
             pagination: Setting::NotSet,
+            embedders: Setting::NotSet,
             _kind: std::marker::PhantomData,
         };
         settings.check()

dump/src/reader/compat/v1_to_v2.rs

@@ -120,7 +120,7 @@ impl From<v1::settings::Settings> for v2::Settings<v2::Unchecked> {
                                 criterion.as_ref().map(ToString::to_string)
                             }
                             Err(()) => {
-                                log::warn!(
+                                tracing::warn!(
                                     "Could not import the following ranking rule: `{}`.",
                                     ranking_rule
                                 );
@@ -152,11 +152,11 @@ impl From<v1::update::UpdateStatus> for Option<v2::updates::UpdateStatus> {
         use v2::updates::UpdateStatus as UpdateStatusV2;
         Some(match source {
             UpdateStatusV1::Enqueued { content } => {
-                log::warn!(
+                tracing::warn!(
                     "Cannot import task {} (importing enqueued tasks from v1 dumps is unsupported)",
                     content.update_id
                 );
-                log::warn!("Task will be skipped in the queue of imported tasks.");
+                tracing::warn!("Task will be skipped in the queue of imported tasks.");
 
                 return None;
             }
@@ -229,7 +229,7 @@ impl From<v1::update::UpdateType> for Option<v2::updates::UpdateMeta> {
         Some(match source {
             v1::update::UpdateType::ClearAll => v2::updates::UpdateMeta::ClearDocuments,
             v1::update::UpdateType::Customs => {
-                log::warn!("Ignoring task with type 'Customs' that is no longer supported");
+                tracing::warn!("Ignoring task with type 'Customs' that is no longer supported");
                 return None;
             }
             v1::update::UpdateType::DocumentsAddition { .. } => {
@@ -296,7 +296,7 @@ impl From<v1::settings::RankingRule> for Option<v2::settings::Criterion> {
             v1::settings::RankingRule::Proximity => Some(v2::settings::Criterion::Proximity),
             v1::settings::RankingRule::Attribute => Some(v2::settings::Criterion::Attribute),
             v1::settings::RankingRule::WordsPosition => {
-                log::warn!("Removing the 'WordsPosition' ranking rule that is no longer supported, please check the resulting ranking rules of your indexes");
+                tracing::warn!("Removing the 'WordsPosition' ranking rule that is no longer supported, please check the resulting ranking rules of your indexes");
                 None
             }
             v1::settings::RankingRule::Exactness => Some(v2::settings::Criterion::Exactness),

dump/src/reader/compat/v2_to_v3.rs

@@ -146,8 +146,8 @@ impl From<v2::updates::UpdateStatus> for v3::updates::UpdateStatus {
                         started_processing_at: processing.started_processing_at,
                     }),
                     Err(e) => {
-                        log::warn!("Error with task {}: {}", processing.from.update_id, e);
-                        log::warn!("Task will be marked as `Failed`.");
+                        tracing::warn!("Error with task {}: {}", processing.from.update_id, e);
+                        tracing::warn!("Task will be marked as `Failed`.");
                         v3::updates::UpdateStatus::Failed(v3::updates::Failed {
                             from: v3::updates::Processing {
                                 from: v3::updates::Enqueued {
@@ -172,8 +172,8 @@ impl From<v2::updates::UpdateStatus> for v3::updates::UpdateStatus {
                         enqueued_at: enqueued.enqueued_at,
                     }),
                     Err(e) => {
-                        log::warn!("Error with task {}: {}", enqueued.update_id, e);
-                        log::warn!("Task will be marked as `Failed`.");
+                        tracing::warn!("Error with task {}: {}", enqueued.update_id, e);
+                        tracing::warn!("Task will be marked as `Failed`.");
                         v3::updates::UpdateStatus::Failed(v3::updates::Failed {
                             from: v3::updates::Processing {
                                 from: v3::updates::Enqueued {
@@ -353,7 +353,7 @@ impl From<String> for v3::Code {
             "malformed_payload" => v3::Code::MalformedPayload,
             "missing_payload" => v3::Code::MissingPayload,
             other => {
-                log::warn!("Unknown error code {}", other);
+                tracing::warn!("Unknown error code {}", other);
                 v3::Code::UnretrievableErrorCode
             }
         }

dump/src/reader/compat/v3_to_v4.rs

@@ -76,20 +76,20 @@ impl CompatV3ToV4 {
                         let index_uid = match index_uid {
                             Some(uid) => uid,
                             None => {
-                                log::warn!(
+                                tracing::warn!(
                                     "Error while importing the update {}.",
                                     task.update.id()
                                 );
-                                log::warn!(
+                                tracing::warn!(
                                     "The index associated to the uuid `{}` could not be retrieved.",
                                     task.uuid.to_string()
                                 );
                                 if task.update.is_finished() {
                                     // we're fucking with his history but not his data, that's ok-ish.
-                                    log::warn!("The index-uuid will be set as `unknown`.");
+                                    tracing::warn!("The index-uuid will be set as `unknown`.");
                                     String::from("unknown")
                                 } else {
-                                    log::warn!("The task will be ignored.");
+                                    tracing::warn!("The task will be ignored.");
                                     return None;
                                 }
                             }

dump/src/reader/compat/v4_to_v5.rs

@@ -305,7 +305,7 @@ impl From<v4::ResponseError> for v5::ResponseError {
             "invalid_api_key_expires_at" => v5::Code::InvalidApiKeyExpiresAt,
             "invalid_api_key_description" => v5::Code::InvalidApiKeyDescription,
             other => {
-                log::warn!("Unknown error code {}", other);
+                tracing::warn!("Unknown error code {}", other);
                 v5::Code::UnretrievableErrorCode
             }
         };

dump/src/reader/compat/v5_to_v6.rs

@@ -304,7 +304,7 @@ impl From<v5::ResponseError> for v6::ResponseError {
             "immutable_field" => v6::Code::BadRequest,
             "api_key_already_exists" => v6::Code::ApiKeyAlreadyExists,
             other => {
-                log::warn!("Unknown error code {}", other);
+                tracing::warn!("Unknown error code {}", other);
                 v6::Code::UnretrievableErrorCode
             }
         };
@@ -329,7 +329,7 @@ impl<T> From<v5::Settings<T>> for v6::Settings<v6::Unchecked> {
                                     new_ranking_rules.push(new_rule);
                                 }
                                 Err(_) => {
-                                    log::warn!("Error while importing settings. The ranking rule `{rule}` does not exist anymore.")
+                                    tracing::warn!("Error while importing settings. The ranking rule `{rule}` does not exist anymore.")
                                 }
                             }
                         }
@@ -378,6 +378,7 @@ impl<T> From<v5::Settings<T>> for v6::Settings<v6::Unchecked> {
                 v5::Setting::Reset => v6::Setting::Reset,
                 v5::Setting::NotSet => v6::Setting::NotSet,
             },
+            embedders: v6::Setting::NotSet,
             _kind: std::marker::PhantomData,
         }
     }

dump/src/reader/v3/errors.rs

@@ -1,5 +1,6 @@
 use serde::{Deserialize, Serialize};
 
+#[allow(clippy::enum_variant_names)]
 #[derive(Serialize, Deserialize, Debug, Clone, Copy)]
 pub enum Code {
     // index related error

dump/src/reader/v4/errors.rs

@@ -95,6 +95,7 @@ impl fmt::Display for ErrorType {
     }
 }
 
+#[allow(clippy::enum_variant_names)]
 #[derive(Serialize, Deserialize, Debug, Clone, Copy)]
 pub enum Code {
     // index related error

dump/src/reader/v5/errors.rs

@@ -31,6 +31,7 @@ impl ResponseError {
     }
 }
 
+#[allow(clippy::enum_variant_names)]
 #[derive(Deserialize, Debug, Clone, Copy)]
 #[cfg_attr(test, derive(serde::Serialize))]
 pub enum Code {

dump/src/reader/v6/mod.rs

@@ -2,10 +2,10 @@ use std::fs::{self, File};
 use std::io::{BufRead, BufReader, ErrorKind};
 use std::path::Path;
 
-use log::debug;
 pub use meilisearch_types::milli;
 use tempfile::TempDir;
 use time::OffsetDateTime;
+use tracing::debug;
 use uuid::Uuid;
 
 use super::Document;

file-store/Cargo.toml

@@ -11,9 +11,9 @@ edition.workspace = true
 license.workspace = true
 
 [dependencies]
-tempfile = "3.5.0"
-thiserror = "1.0.40"
-uuid = { version = "1.3.1", features = ["serde", "v4"] }
+tempfile = "3.9.0"
+thiserror = "1.0.56"
+uuid = { version = "1.6.1", features = ["serde", "v4"] }
 
 [dev-dependencies]
-faux = "0.1.9"
+faux = "0.1.10"

filter-parser/Cargo.toml

@@ -13,8 +13,8 @@ license.workspace = true
 
 [dependencies]
 nom = "7.1.3"
-nom_locate = "4.1.0"
-unescaper = "0.1.2"
+nom_locate = "4.2.0"
+unescaper = "0.1.3"
 
 [dev-dependencies]
-insta = "1.29.0"
+insta = "1.34.0"

fuzzers/Cargo.toml

@@ -11,10 +11,10 @@ edition.workspace = true
 license.workspace = true
 
 [dependencies]
-arbitrary = { version = "1.3.0", features = ["derive"] }
-clap = { version = "4.3.0", features = ["derive"] }
-fastrand = "2.0.0"
+arbitrary = { version = "1.3.2", features = ["derive"] }
+clap = { version = "4.4.17", features = ["derive"] }
+fastrand = "2.0.1"
 milli = { path = "../milli" }
-serde = { version = "1.0.160", features = ["derive"] }
-serde_json = { version = "1.0.95", features = ["preserve_order"] }
-tempfile = "3.5.0"
+serde = { version = "1.0.195", features = ["derive"] }
+serde_json = { version = "1.0.111", features = ["preserve_order"] }
+tempfile = "3.9.0"

index-scheduler/Cargo.toml

@@ -11,30 +11,36 @@ edition.workspace = true
 license.workspace = true
 
 [dependencies]
-anyhow = "1.0.70"
+anyhow = "1.0.79"
 bincode = "1.3.3"
-csv = "1.2.1"
+csv = "1.3.0"
 derive_builder = "0.12.0"
 dump = { path = "../dump" }
-enum-iterator = "1.4.0"
+enum-iterator = "1.5.0"
 file-store = { path = "../file-store" }
-log = "0.4.17"
+flate2 = "1.0.28"
 meilisearch-auth = { path = "../meilisearch-auth" }
 meilisearch-types = { path = "../meilisearch-types" }
 page_size = "0.5.0"
 puffin = { version = "0.16.0", features = ["serialization"] }
-roaring = { version = "0.10.1", features = ["serde"] }
-serde = { version = "1.0.160", features = ["derive"] }
-serde_json = { version = "1.0.95", features = ["preserve_order"] }
+roaring = { version = "0.10.2", features = ["serde"] }
+serde = { version = "1.0.195", features = ["derive"] }
+serde_json = { version = "1.0.111", features = ["preserve_order"] }
 synchronoise = "1.0.1"
-tempfile = "3.5.0"
-thiserror = "1.0.40"
-time = { version = "0.3.20", features = ["serde-well-known", "formatting", "parsing", "macros"] }
-uuid = { version = "1.3.1", features = ["serde", "v4"] }
+tempfile = "3.9.0"
+thiserror = "1.0.56"
+time = { version = "0.3.31", features = [
+    "serde-well-known",
+    "formatting",
+    "parsing",
+    "macros",
+] }
+tracing = "0.1.40"
+ureq = "2.9.1"
+uuid = { version = "1.6.1", features = ["serde", "v4"] }
 
 [dev-dependencies]
 big_s = "1.0.2"
-crossbeam = "0.8.2"
-insta = { version = "1.29.0", features = ["json", "redactions"] }
+crossbeam = "0.8.4"
+insta = { version = "1.34.0", features = ["json", "redactions"] }
 meili-snap = { path = "../meili-snap" }
-nelson = { git = "https://github.com/meilisearch/nelson.git", rev = "675f13885548fb415ead8fbb447e9e6d9314000a"}

index-scheduler/src/batch.rs

@@ -24,7 +24,6 @@ use std::fs::{self, File};
 use std::io::BufWriter;
 
 use dump::IndexMetadata;
-use log::{debug, error, info, trace};
 use meilisearch_types::error::Code;
 use meilisearch_types::heed::{RoTxn, RwTxn};
 use meilisearch_types::milli::documents::{obkv_to_object, DocumentsBatchReader};
@@ -60,7 +59,7 @@ pub(crate) enum Batch {
         /// The list of tasks that were processing when this task cancelation appeared.
         previous_processing_tasks: RoaringBitmap,
     },
-    TaskDeletion(Task),
+    TaskDeletions(Vec<Task>),
     SnapshotCreation(Vec<Task>),
     Dump(Task),
     IndexOperation {
@@ -146,13 +145,12 @@ impl Batch {
     pub fn ids(&self) -> Vec<TaskId> {
         match self {
             Batch::TaskCancelation { task, .. }
-            | Batch::TaskDeletion(task)
             | Batch::Dump(task)
             | Batch::IndexCreation { task, .. }
             | Batch::IndexUpdate { task, .. } => vec![task.uid],
-            Batch::SnapshotCreation(tasks) | Batch::IndexDeletion { tasks, .. } => {
-                tasks.iter().map(|task| task.uid).collect()
-            }
+            Batch::SnapshotCreation(tasks)
+            | Batch::TaskDeletions(tasks)
+            | Batch::IndexDeletion { tasks, .. } => tasks.iter().map(|task| task.uid).collect(),
             Batch::IndexOperation { op, .. } => match op {
                 IndexOperation::DocumentOperation { tasks, .. }
                 | IndexOperation::Settings { tasks, .. }
@@ -180,7 +178,7 @@ impl Batch {
         use Batch::*;
         match self {
             TaskCancelation { .. }
-            | TaskDeletion(_)
+            | TaskDeletions(_)
             | SnapshotCreation(_)
             | Dump(_)
             | IndexSwap { .. } => None,
@@ -199,7 +197,7 @@ impl fmt::Display for Batch {
         let tasks = self.ids();
         match self {
             Batch::TaskCancelation { .. } => f.write_str("TaskCancelation")?,
-            Batch::TaskDeletion(_) => f.write_str("TaskDeletion")?,
+            Batch::TaskDeletions(_) => f.write_str("TaskDeletion")?,
             Batch::SnapshotCreation(_) => f.write_str("SnapshotCreation")?,
             Batch::Dump(_) => f.write_str("Dump")?,
             Batch::IndexOperation { op, .. } => write!(f, "{op}")?,
@@ -515,6 +513,7 @@ impl IndexScheduler {
     /// 3. We get the *next* snapshot to process.
     /// 4. We get the *next* dump to process.
     /// 5. We get the *next* tasks to process for a specific index.
+    #[tracing::instrument(level = "trace", skip(self, rtxn), target = "indexing::scheduler")]
     pub(crate) fn create_next_batch(&self, rtxn: &RoTxn) -> Result<Option<Batch>> {
         #[cfg(test)]
         self.maybe_fail(crate::tests::FailureLocation::InsideCreateBatch)?;
@@ -539,9 +538,9 @@ impl IndexScheduler {
 
         // 2. we get the next task to delete
         let to_delete = self.get_kind(rtxn, Kind::TaskDeletion)? & enqueued;
-        if let Some(task_id) = to_delete.min() {
-            let task = self.get_task(rtxn, task_id)?.ok_or(Error::CorruptedTaskQueue)?;
-            return Ok(Some(Batch::TaskDeletion(task)));
+        if !to_delete.is_empty() {
+            let tasks = self.get_existing_tasks(rtxn, to_delete)?;
+            return Ok(Some(Batch::TaskDeletions(tasks)));
         }
 
         // 3. we batch the snapshot.
@@ -584,7 +583,9 @@ impl IndexScheduler {
         let index_tasks = self.index_tasks(rtxn, index_name)? & enqueued;
 
         // If autobatching is disabled we only take one task at a time.
-        let tasks_limit = if self.autobatching_enabled { usize::MAX } else { 1 };
+        // Otherwise, we take only a maximum of tasks to create batches.
+        let tasks_limit =
+            if self.autobatching_enabled { self.max_number_of_batched_tasks } else { 1 };
 
         let enqueued = index_tasks
             .into_iter()
@@ -618,6 +619,7 @@ impl IndexScheduler {
     /// The list of tasks that were processed. The metadata of each task in the returned
     /// list is updated accordingly, with the exception of the its date fields
     /// [`finished_at`](meilisearch_types::tasks::Task::finished_at) and [`started_at`](meilisearch_types::tasks::Task::started_at).
+    #[tracing::instrument(level = "trace", skip(self, batch), target = "indexing::scheduler", fields(batch=batch.to_string()))]
     pub(crate) fn process_batch(&self, batch: Batch) -> Result<Vec<Task>> {
         #[cfg(test)]
         {
@@ -667,9 +669,10 @@ impl IndexScheduler {
                     Ok(()) => {
                         for content_uuid in canceled_tasks_content_uuids {
                             if let Err(error) = self.delete_update_file(content_uuid) {
-                                error!(
-                                    "We failed deleting the content file indentified as {}: {}",
-                                    content_uuid, error
+                                tracing::error!(
+                                    file_content_uuid = %content_uuid,
+                                    %error,
+                                    "Failed deleting content file"
                                 )
                             }
                         }
@@ -679,31 +682,43 @@ impl IndexScheduler {
 
                 Ok(vec![task])
             }
-            Batch::TaskDeletion(mut task) => {
+            Batch::TaskDeletions(mut tasks) => {
                 // 1. Retrieve the tasks that matched the query at enqueue-time.
-                let matched_tasks =
+                let mut matched_tasks = RoaringBitmap::new();
+
+                for task in tasks.iter() {
                     if let KindWithContent::TaskDeletion { tasks, query: _ } = &task.kind {
-                        tasks
+                        matched_tasks |= tasks;
                     } else {
                         unreachable!()
+                    }
+                }
+
+                let mut wtxn = self.env.write_txn()?;
+                let mut deleted_tasks = self.delete_matched_tasks(&mut wtxn, &matched_tasks)?;
+                wtxn.commit()?;
+
+                for task in tasks.iter_mut() {
+                    task.status = Status::Succeeded;
+                    let KindWithContent::TaskDeletion { tasks, query: _ } = &task.kind else {
+                        unreachable!()
                     };
 
-                let mut wtxn = self.env.write_txn()?;
-                let deleted_tasks_count = self.delete_matched_tasks(&mut wtxn, matched_tasks)?;
+                    let deleted_tasks_count = deleted_tasks.intersection_len(tasks);
+                    deleted_tasks -= tasks;
 
-                task.status = Status::Succeeded;
-                match &mut task.details {
-                    Some(Details::TaskDeletion {
-                        matched_tasks: _,
-                        deleted_tasks,
-                        original_filter: _,
-                    }) => {
-                        *deleted_tasks = Some(deleted_tasks_count);
+                    match &mut task.details {
+                        Some(Details::TaskDeletion {
+                            matched_tasks: _,
+                            deleted_tasks,
+                            original_filter: _,
+                        }) => {
+                            *deleted_tasks = Some(deleted_tasks_count);
+                        }
+                        _ => unreachable!(),
                     }
-                    _ => unreachable!(),
                 }
-                wtxn.commit()?;
-                Ok(vec![task])
+                Ok(tasks)
             }
             Batch::SnapshotCreation(mut tasks) => {
                 fs::create_dir_all(&self.snapshots_path)?;
@@ -934,8 +949,8 @@ impl IndexScheduler {
                 };
 
                 // the index operation can take a long time, so save this handle to make it available to the search for the duration of the tick
-                *self.currently_updating_index.write().unwrap() =
-                    Some((index_uid.clone(), index.clone()));
+                self.index_mapper
+                    .set_currently_updating_index(Some((index_uid.clone(), index.clone())));
 
                 let mut index_wtxn = index.write_txn()?;
                 let tasks = self.apply_index_operation(&mut index_wtxn, &index, op)?;
@@ -956,7 +971,10 @@ impl IndexScheduler {
 
                 match res {
                     Ok(_) => (),
-                    Err(e) => error!("Could not write the stats of the index {}", e),
+                    Err(e) => tracing::error!(
+                        error = &e as &dyn std::error::Error,
+                        "Could not write the stats of the index"
+                    ),
                 }
 
                 Ok(tasks)
@@ -984,7 +1002,7 @@ impl IndexScheduler {
                     builder.set_primary_key(primary_key);
                     let must_stop_processing = self.must_stop_processing.clone();
                     builder.execute(
-                        |indexing_step| debug!("update: {:?}", indexing_step),
+                        |indexing_step| tracing::debug!(update = ?indexing_step),
                         || must_stop_processing.get(),
                     )?;
                     index_wtxn.commit()?;
@@ -1011,7 +1029,10 @@ impl IndexScheduler {
 
                 match res {
                     Ok(_) => (),
-                    Err(e) => error!("Could not write the stats of the index {}", e),
+                    Err(e) => tracing::error!(
+                        error = &e as &dyn std::error::Error,
+                        "Could not write the stats of the index"
+                    ),
                 }
 
                 Ok(vec![task])
@@ -1130,6 +1151,11 @@ impl IndexScheduler {
     ///
     /// ## Return
     /// The list of processed tasks.
+    #[tracing::instrument(
+        level = "trace",
+        skip(self, index_wtxn, index),
+        target = "indexing::scheduler"
+    )]
     fn apply_index_operation<'i>(
         &self,
         index_wtxn: &mut RwTxn<'i>,
@@ -1190,7 +1216,7 @@ impl IndexScheduler {
                                 milli::update::Settings::new(index_wtxn, index, indexer_config);
                             builder.set_primary_key(primary_key);
                             builder.execute(
-                                |indexing_step| debug!("update: {:?}", indexing_step),
+                                |indexing_step| tracing::debug!(update = ?indexing_step),
                                 || must_stop_processing.clone().get(),
                             )?;
                             primary_key_has_been_set = true;
@@ -1200,12 +1226,16 @@ impl IndexScheduler {
 
                 let config = IndexDocumentsConfig { update_method: method, ..Default::default() };
 
+                let embedder_configs = index.embedding_configs(index_wtxn)?;
+                // TODO: consider Arc'ing the map too (we only need read access + we'll be cloning it multiple times, so really makes sense)
+                let embedders = self.embedders(embedder_configs)?;
+
                 let mut builder = milli::update::IndexDocuments::new(
                     index_wtxn,
                     index,
                     indexer_config,
                     config,
-                    |indexing_step| trace!("update: {:?}", indexing_step),
+                    |indexing_step| tracing::trace!(?indexing_step, "Update"),
                     || must_stop_processing.get(),
                 )?;
 
@@ -1218,6 +1248,8 @@ impl IndexScheduler {
                             let (new_builder, user_result) = builder.add_documents(reader)?;
                             builder = new_builder;
 
+                            builder = builder.with_embedders(embedders.clone());
+
                             let received_documents =
                                 if let Some(Details::DocumentAdditionOrUpdate {
                                     received_documents,
@@ -1275,7 +1307,7 @@ impl IndexScheduler {
 
                 if !tasks.iter().all(|res| res.error.is_some()) {
                     let addition = builder.execute()?;
-                    info!("document addition done: {:?}", addition);
+                    tracing::info!(indexing_result = ?addition, "document indexing done");
                 } else if primary_key_has_been_set {
                     // Everything failed but we've set a primary key.
                     // We need to remove it.
@@ -1283,7 +1315,7 @@ impl IndexScheduler {
                         milli::update::Settings::new(index_wtxn, index, indexer_config);
                     builder.reset_primary_key();
                     builder.execute(
-                        |indexing_step| trace!("update: {:?}", indexing_step),
+                        |indexing_step| tracing::trace!(update = ?indexing_step),
                         || must_stop_processing.clone().get(),
                     )?;
                 }
@@ -1343,9 +1375,6 @@ impl IndexScheduler {
 
                 for (task, (_, settings)) in tasks.iter_mut().zip(settings) {
                     let checked_settings = settings.clone().check();
-                    if checked_settings.proximity_precision.set().is_some() {
-                        self.features.features().check_proximity_precision()?;
-                    }
                     task.details = Some(Details::SettingsUpdate { settings: Box::new(settings) });
                     apply_settings_to_builder(&checked_settings, &mut builder);
 
@@ -1356,7 +1385,7 @@ impl IndexScheduler {
 
                 let must_stop_processing = self.must_stop_processing.clone();
                 builder.execute(
-                    |indexing_step| debug!("update: {:?}", indexing_step),
+                    |indexing_step| tracing::debug!(update = ?indexing_step),
                     || must_stop_processing.get(),
                 )?;
 
@@ -1430,7 +1459,11 @@ impl IndexScheduler {
     /// Delete each given task from all the databases (if it is deleteable).
     ///
     /// Return the number of tasks that were actually deleted.
-    fn delete_matched_tasks(&self, wtxn: &mut RwTxn, matched_tasks: &RoaringBitmap) -> Result<u64> {
+    fn delete_matched_tasks(
+        &self,
+        wtxn: &mut RwTxn,
+        matched_tasks: &RoaringBitmap,
+    ) -> Result<RoaringBitmap> {
         // 1. Remove from this list the tasks that we are not allowed to delete
         let enqueued_tasks = self.get_status(wtxn, Status::Enqueued)?;
         let processing_tasks = &self.processing_tasks.read().unwrap().processing.clone();
@@ -1495,7 +1528,7 @@ impl IndexScheduler {
             }
         }
 
-        Ok(to_delete_tasks.len())
+        Ok(to_delete_tasks)
     }
 
     /// Cancel each given task from all the databases (if it is cancelable).
@@ -1564,7 +1597,7 @@ fn delete_document_by_filter<'a>(
             index,
             indexer_config,
             config,
-            |indexing_step| debug!("update: {:?}", indexing_step),
+            |indexing_step| tracing::debug!(update = ?indexing_step),
             || must_stop_processing.get(),
         )?;
 

index-scheduler/src/features.rs

@@ -30,19 +30,6 @@ impl RoFeatures {
         self.runtime
     }
 
-    pub fn check_score_details(&self) -> Result<()> {
-        if self.runtime.score_details {
-            Ok(())
-        } else {
-            Err(FeatureNotEnabledError {
-                disabled_action: "Computing score details",
-                feature: "score details",
-                issue_link: "https://github.com/meilisearch/product/discussions/674",
-            }
-            .into())
-        }
-    }
-
     pub fn check_metrics(&self) -> Result<()> {
         if self.runtime.metrics {
             Ok(())
@@ -56,12 +43,25 @@ impl RoFeatures {
         }
     }
 
-    pub fn check_vector(&self) -> Result<()> {
+    pub fn check_logs_route(&self) -> Result<()> {
+        if self.runtime.logs_route {
+            Ok(())
+        } else {
+            Err(FeatureNotEnabledError {
+                disabled_action: "getting logs through the `/logs/stream` route",
+                feature: "logs route",
+                issue_link: "https://github.com/orgs/meilisearch/discussions/721",
+            }
+            .into())
+        }
+    }
+
+    pub fn check_vector(&self, disabled_action: &'static str) -> Result<()> {
         if self.runtime.vector_store {
             Ok(())
         } else {
             Err(FeatureNotEnabledError {
-                disabled_action: "Passing `vector` as a query parameter",
+                disabled_action,
                 feature: "vector store",
                 issue_link: "https://github.com/meilisearch/product/discussions/677",
             }
@@ -81,19 +81,6 @@ impl RoFeatures {
             .into())
         }
     }
-
-    pub fn check_proximity_precision(&self) -> Result<()> {
-        if self.runtime.proximity_precision {
-            Ok(())
-        } else {
-            Err(FeatureNotEnabledError {
-                disabled_action: "Using `proximityPrecision` index setting",
-                feature: "proximity precision",
-                issue_link: "https://github.com/orgs/meilisearch/discussions/710",
-            }
-            .into())
-        }
-    }
 }
 
 impl FeatureData {
@@ -107,6 +94,7 @@ impl FeatureData {
             runtime_features_db.get(&txn, EXPERIMENTAL_FEATURES)?.unwrap_or_default();
         let runtime = Arc::new(RwLock::new(RuntimeTogglableFeatures {
             metrics: instance_features.metrics || persisted_features.metrics,
+            logs_route: instance_features.logs_route || persisted_features.logs_route,
             ..persisted_features
         }));
 

index-scheduler/src/index_mapper/mod.rs

@@ -3,13 +3,13 @@ use std::sync::{Arc, RwLock};
 use std::time::Duration;
 use std::{fs, thread};
 
-use log::error;
 use meilisearch_types::heed::types::{SerdeJson, Str};
 use meilisearch_types::heed::{Database, Env, RoTxn, RwTxn};
 use meilisearch_types::milli::update::IndexerConfig;
 use meilisearch_types::milli::{FieldDistribution, Index};
 use serde::{Deserialize, Serialize};
 use time::OffsetDateTime;
+use tracing::error;
 use uuid::Uuid;
 
 use self::index_map::IndexMap;
@@ -69,6 +69,10 @@ pub struct IndexMapper {
     /// Whether we open a meilisearch index with the MDB_WRITEMAP option or not.
     enable_mdb_writemap: bool,
     pub indexer_config: Arc<IndexerConfig>,
+
+    /// A few types of long running batches of tasks that act on a single index set this field
+    /// so that a handle to the index is available from other threads (search) in an optimized manner.
+    currently_updating_index: Arc<RwLock<Option<(String, Index)>>>,
 }
 
 /// Whether the index is available for use or is forbidden to be inserted back in the index map
@@ -151,6 +155,7 @@ impl IndexMapper {
             index_growth_amount,
             enable_mdb_writemap,
             indexer_config: Arc::new(indexer_config),
+            currently_updating_index: Default::default(),
         })
     }
 
@@ -303,6 +308,14 @@ impl IndexMapper {
 
     /// Return an index, may open it if it wasn't already opened.
     pub fn index(&self, rtxn: &RoTxn, name: &str) -> Result<Index> {
+        if let Some((current_name, current_index)) =
+            self.currently_updating_index.read().unwrap().as_ref()
+        {
+            if current_name == name {
+                return Ok(current_index.clone());
+            }
+        }
+
         let uuid = self
             .index_mapping
             .get(rtxn, name)?
@@ -474,4 +487,8 @@ impl IndexMapper {
     pub fn indexer_config(&self) -> &IndexerConfig {
         &self.indexer_config
     }
+
+    pub fn set_currently_updating_index(&self, index: Option<(String, Index)>) {
+        *self.currently_updating_index.write().unwrap() = index;
+    }
 }

index-scheduler/src/insta_snapshot.rs

@@ -30,16 +30,19 @@ pub fn snapshot_index_scheduler(scheduler: &IndexScheduler) -> String {
         index_mapper,
         features: _,
         max_number_of_tasks: _,
+        max_number_of_batched_tasks: _,
         puffin_frame: _,
         wake_up: _,
         dumps_path: _,
         snapshots_path: _,
         auth_path: _,
         version_file_path: _,
+        webhook_url: _,
+        webhook_authorization_header: _,
         test_breakpoint_sdr: _,
         planned_failures: _,
         run_loop_iteration: _,
-        currently_updating_index: _,
+        embedders: _,
     } = scheduler;
 
     let rtxn = env.read_txn().unwrap();

index-scheduler/src/lib.rs

@@ -34,6 +34,7 @@ pub type TaskId = u32;
 
 use std::collections::{BTreeMap, HashMap};
 use std::fs::File;
+use std::io::{self, BufReader, Read};
 use std::ops::{Bound, RangeBounds};
 use std::path::{Path, PathBuf};
 use std::sync::atomic::AtomicBool;
@@ -45,6 +46,8 @@ use dump::{KindDump, TaskDump, UpdateFile};
 pub use error::Error;
 pub use features::RoFeatures;
 use file_store::FileStore;
+use flate2::bufread::GzEncoder;
+use flate2::Compression;
 use meilisearch_types::error::ResponseError;
 use meilisearch_types::features::{InstanceTogglableFeatures, RuntimeTogglableFeatures};
 use meilisearch_types::heed::byteorder::BE;
@@ -52,7 +55,9 @@ use meilisearch_types::heed::types::{SerdeBincode, SerdeJson, Str, I128};
 use meilisearch_types::heed::{self, Database, Env, PutFlags, RoTxn, RwTxn};
 use meilisearch_types::milli::documents::DocumentsBatchBuilder;
 use meilisearch_types::milli::update::IndexerConfig;
+use meilisearch_types::milli::vector::{Embedder, EmbedderOptions, EmbeddingConfigs};
 use meilisearch_types::milli::{self, CboRoaringBitmapCodec, Index, RoaringBitmapCodec, BEU32};
+use meilisearch_types::task_view::TaskView;
 use meilisearch_types::tasks::{Kind, KindWithContent, Status, Task};
 use puffin::FrameView;
 use roaring::RoaringBitmap;
@@ -169,8 +174,8 @@ impl ProcessingTasks {
     }
 
     /// Set the processing tasks to an empty list
-    fn stop_processing(&mut self) {
-        self.processing = RoaringBitmap::new();
+    fn stop_processing(&mut self) -> RoaringBitmap {
+        std::mem::take(&mut self.processing)
     }
 
     /// Returns `true` if there, at least, is one task that is currently processing that we must stop.
@@ -240,6 +245,10 @@ pub struct IndexSchedulerOptions {
     pub snapshots_path: PathBuf,
     /// The path to the folder containing the dumps.
     pub dumps_path: PathBuf,
+    /// The URL on which we must send the tasks statuses
+    pub webhook_url: Option<String>,
+    /// The value we will send into the Authorization HTTP header on the webhook URL
+    pub webhook_authorization_header: Option<String>,
     /// The maximum size, in bytes, of the task index.
     pub task_db_size: usize,
     /// The size, in bytes, with which a meilisearch index is opened the first time of each meilisearch index.
@@ -258,6 +267,9 @@ pub struct IndexSchedulerOptions {
     /// The maximum number of tasks stored in the task queue before starting
     /// to auto schedule task deletions.
     pub max_number_of_tasks: usize,
+    /// If the autobatcher is allowed to automatically batch tasks
+    /// it will only batch this defined number of tasks at once.
+    pub max_number_of_batched_tasks: usize,
     /// The experimental features enabled for this instance.
     pub instance_features: InstanceTogglableFeatures,
 }
@@ -316,6 +328,14 @@ pub struct IndexScheduler {
     /// the finished tasks automatically.
     pub(crate) max_number_of_tasks: usize,
 
+    /// The maximum number of tasks that will be batched together.
+    pub(crate) max_number_of_batched_tasks: usize,
+
+    /// The webhook url we should send tasks to after processing every batches.
+    pub(crate) webhook_url: Option<String>,
+    /// The Authorization header to send to the webhook URL.
+    pub(crate) webhook_authorization_header: Option<String>,
+
     /// A frame to output the indexation profiling files to disk.
     pub(crate) puffin_frame: Arc<puffin::GlobalFrameView>,
 
@@ -331,9 +351,7 @@ pub struct IndexScheduler {
     /// The path to the version file of Meilisearch.
     pub(crate) version_file_path: PathBuf,
 
-    /// A few types of long running batches of tasks that act on a single index set this field
-    /// so that a handle to the index is available from other threads (search) in an optimized manner.
-    currently_updating_index: Arc<RwLock<Option<(String, Index)>>>,
+    embedders: Arc<RwLock<HashMap<EmbedderOptions, Arc<Embedder>>>>,
 
     // ================= test
     // The next entry is dedicated to the tests.
@@ -373,12 +391,15 @@ impl IndexScheduler {
             wake_up: self.wake_up.clone(),
             autobatching_enabled: self.autobatching_enabled,
             max_number_of_tasks: self.max_number_of_tasks,
+            max_number_of_batched_tasks: self.max_number_of_batched_tasks,
             puffin_frame: self.puffin_frame.clone(),
             snapshots_path: self.snapshots_path.clone(),
             dumps_path: self.dumps_path.clone(),
             auth_path: self.auth_path.clone(),
             version_file_path: self.version_file_path.clone(),
-            currently_updating_index: self.currently_updating_index.clone(),
+            webhook_url: self.webhook_url.clone(),
+            webhook_authorization_header: self.webhook_authorization_header.clone(),
+            embedders: self.embedders.clone(),
             #[cfg(test)]
             test_breakpoint_sdr: self.test_breakpoint_sdr.clone(),
             #[cfg(test)]
@@ -471,11 +492,14 @@ impl IndexScheduler {
             puffin_frame: Arc::new(puffin::GlobalFrameView::default()),
             autobatching_enabled: options.autobatching_enabled,
             max_number_of_tasks: options.max_number_of_tasks,
+            max_number_of_batched_tasks: options.max_number_of_batched_tasks,
             dumps_path: options.dumps_path,
             snapshots_path: options.snapshots_path,
             auth_path: options.auth_path,
             version_file_path: options.version_file_path,
-            currently_updating_index: Arc::new(RwLock::new(None)),
+            webhook_url: options.webhook_url,
+            webhook_authorization_header: options.webhook_authorization_header,
+            embedders: Default::default(),
 
             #[cfg(test)]
             test_breakpoint_sdr,
@@ -511,17 +535,17 @@ impl IndexScheduler {
         let budget = if Self::is_good_heed(tasks_path, DEFAULT_BUDGET) {
             DEFAULT_BUDGET
         } else {
-            log::debug!("determining budget with dichotomic search");
+            tracing::debug!("determining budget with dichotomic search");
             utils::dichotomic_search(DEFAULT_BUDGET / 2, |map_size| {
                 Self::is_good_heed(tasks_path, map_size)
             })
         };
 
-        log::debug!("memmap budget: {budget}B");
+        tracing::debug!("memmap budget: {budget}B");
         let mut budget = budget / 2;
         if task_db_size > (budget / 2) {
             task_db_size = clamp_to_page_size(budget * 2 / 5);
-            log::debug!(
+            tracing::debug!(
                 "Decreasing max size of task DB to {task_db_size}B due to constrained memory space"
             );
         }
@@ -531,13 +555,13 @@ impl IndexScheduler {
         let budget = budget;
         let task_db_size = task_db_size;
 
-        log::debug!("index budget: {budget}B");
+        tracing::debug!("index budget: {budget}B");
         let mut index_count = budget / base_map_size;
         if index_count < 2 {
             // take a bit less than half than the budget to make sure we can always afford to open an index
             let map_size = (budget * 2) / 5;
             // single index of max budget
-            log::debug!("1 index of {map_size}B can be opened simultaneously.");
+            tracing::debug!("1 index of {map_size}B can be opened simultaneously.");
             return IndexBudget { map_size, index_count: 1, task_db_size };
         }
         // give us some space for an additional index when the cache is already full
@@ -546,7 +570,7 @@ impl IndexScheduler {
         if index_count > max_index_count {
             index_count = max_index_count;
         }
-        log::debug!("Up to {index_count} indexes of {base_map_size}B opened simultaneously.");
+        tracing::debug!("Up to {index_count} indexes of {base_map_size}B opened simultaneously.");
         IndexBudget { map_size: base_map_size, index_count, task_db_size }
     }
 
@@ -593,7 +617,7 @@ impl IndexScheduler {
                         Ok(TickOutcome::TickAgain(_)) => (),
                         Ok(TickOutcome::WaitForSignal) => run.wake_up.wait(),
                         Err(e) => {
-                            log::error!("{e}");
+                            tracing::error!("{e}");
                             // Wait one second when an irrecoverable error occurs.
                             if !e.is_recoverable() {
                                 std::thread::sleep(Duration::from_secs(1));
@@ -610,15 +634,15 @@ impl IndexScheduler {
                             let mut file = match File::create(format!("{}.puffin", now)) {
                                 Ok(file) => file,
                                 Err(e) => {
-                                    log::error!("{e}");
+                                    tracing::error!("{e}");
                                     continue;
                                 }
                             };
                             if let Err(e) = frame_view.save_to_writer(&mut file) {
-                                log::error!("{e}");
+                                tracing::error!("{e}");
                             }
                             if let Err(e) = file.sync_all() {
-                                log::error!("{e}");
+                                tracing::error!("{e}");
                             }
                             // We erase this frame view as it is no more useful. We want to
                             // measure the new frames now that we exported the previous ones.
@@ -658,13 +682,6 @@ impl IndexScheduler {
     /// If you need to fetch information from or perform an action on all indexes,
     /// see the `try_for_each_index` function.
     pub fn index(&self, name: &str) -> Result<Index> {
-        if let Some((current_name, current_index)) =
-            self.currently_updating_index.read().unwrap().as_ref()
-        {
-            if current_name == name {
-                return Ok(current_index.clone());
-            }
-        }
         let rtxn = self.env.read_txn()?;
         self.index_mapper.index(&rtxn, name)
     }
@@ -1145,7 +1162,7 @@ impl IndexScheduler {
         };
 
         // Reset the currently updating index to relinquish the index handle
-        *self.currently_updating_index.write().unwrap() = None;
+        self.index_mapper.set_currently_updating_index(None);
 
         #[cfg(test)]
         self.maybe_fail(tests::FailureLocation::AcquiringWtxn)?;
@@ -1173,10 +1190,10 @@ impl IndexScheduler {
                     self.update_task(&mut wtxn, &task)
                         .map_err(|e| Error::TaskDatabaseUpdate(Box::new(e)))?;
                     if let Err(e) = self.delete_persisted_task_data(&task) {
-                        log::error!("Failure to delete the content files associated with task {}. Error: {e}", task.uid);
+                        tracing::error!("Failure to delete the content files associated with task {}. Error: {e}", task.uid);
                     }
                 }
-                log::info!("A batch of tasks was successfully completed.");
+                tracing::info!("A batch of tasks was successfully completed.");
             }
             // If we have an abortion error we must stop the tick here and re-schedule tasks.
             Err(Error::Milli(milli::Error::InternalError(
@@ -1230,7 +1247,7 @@ impl IndexScheduler {
                     self.maybe_fail(tests::FailureLocation::UpdatingTaskAfterProcessBatchFailure)?;
 
                     if let Err(e) = self.delete_persisted_task_data(&task) {
-                        log::error!("Failure to delete the content files associated with task {}. Error: {e}", task.uid);
+                        tracing::error!("Failure to delete the content files associated with task {}. Error: {e}", task.uid);
                     }
                     self.update_task(&mut wtxn, &task)
                         .map_err(|e| Error::TaskDatabaseUpdate(Box::new(e)))?;
@@ -1238,19 +1255,99 @@ impl IndexScheduler {
             }
         }
 
-        self.processing_tasks.write().unwrap().stop_processing();
+        let processed = self.processing_tasks.write().unwrap().stop_processing();
 
         #[cfg(test)]
         self.maybe_fail(tests::FailureLocation::CommittingWtxn)?;
 
         wtxn.commit().map_err(Error::HeedTransaction)?;
 
+        // We shouldn't crash the tick function if we can't send data to the webhook.
+        let _ = self.notify_webhook(&processed);
+
         #[cfg(test)]
         self.breakpoint(Breakpoint::AfterProcessing);
 
         Ok(TickOutcome::TickAgain(processed_tasks))
     }
 
+    /// Once the tasks changes have been commited we must send all the tasks that were updated to our webhook if there is one.
+    fn notify_webhook(&self, updated: &RoaringBitmap) -> Result<()> {
+        if let Some(ref url) = self.webhook_url {
+            struct TaskReader<'a, 'b> {
+                rtxn: &'a RoTxn<'a>,
+                index_scheduler: &'a IndexScheduler,
+                tasks: &'b mut roaring::bitmap::Iter<'b>,
+                buffer: Vec<u8>,
+                written: usize,
+            }
+
+            impl<'a, 'b> Read for TaskReader<'a, 'b> {
+                fn read(&mut self, mut buf: &mut [u8]) -> std::io::Result<usize> {
+                    if self.buffer.is_empty() {
+                        match self.tasks.next() {
+                            None => return Ok(0),
+                            Some(task_id) => {
+                                let task = self
+                                    .index_scheduler
+                                    .get_task(self.rtxn, task_id)
+                                    .map_err(|err| io::Error::new(io::ErrorKind::Other, err))?
+                                    .ok_or_else(|| {
+                                        io::Error::new(
+                                            io::ErrorKind::Other,
+                                            Error::CorruptedTaskQueue,
+                                        )
+                                    })?;
+
+                                serde_json::to_writer(
+                                    &mut self.buffer,
+                                    &TaskView::from_task(&task),
+                                )?;
+                                self.buffer.push(b'\n');
+                            }
+                        }
+                    }
+
+                    let mut to_write = &self.buffer[self.written..];
+                    let wrote = io::copy(&mut to_write, &mut buf)?;
+                    self.written += wrote as usize;
+
+                    // we wrote everything and must refresh our buffer on the next call
+                    if self.written == self.buffer.len() {
+                        self.written = 0;
+                        self.buffer.clear();
+                    }
+
+                    Ok(wrote as usize)
+                }
+            }
+
+            let rtxn = self.env.read_txn()?;
+
+            let task_reader = TaskReader {
+                rtxn: &rtxn,
+                index_scheduler: self,
+                tasks: &mut updated.into_iter(),
+                buffer: Vec::with_capacity(50), // on average a task is around ~100 bytes
+                written: 0,
+            };
+
+            // let reader = GzEncoder::new(BufReader::new(task_reader), Compression::default());
+            let reader = GzEncoder::new(BufReader::new(task_reader), Compression::default());
+            let request = ureq::post(url).set("Content-Encoding", "gzip");
+            let request = match &self.webhook_authorization_header {
+                Some(header) => request.set("Authorization", header),
+                None => request,
+            };
+
+            if let Err(e) = request.send(reader) {
+                tracing::error!("While sending data to the webhook: {e}");
+            }
+        }
+
+        Ok(())
+    }
+
     /// Register a task to cleanup the task queue if needed
     fn cleanup_task_queue(&self) -> Result<()> {
         let rtxn = self.env.read_txn().map_err(Error::HeedTransaction)?;
@@ -1270,12 +1367,12 @@ impl IndexScheduler {
         // /!\ the len must be at least 2 or else we might enter an infinite loop where we only delete
         //     the deletion tasks we enqueued ourselves.
         if to_delete.len() < 2 {
-            log::warn!("The task queue is almost full, but no task can be deleted yet.");
+            tracing::warn!("The task queue is almost full, but no task can be deleted yet.");
             // the only thing we can do is hope that the user tasks are going to finish
             return Ok(());
         }
 
-        log::info!(
+        tracing::info!(
             "The task queue is almost full. Deleting the oldest {} finished tasks.",
             to_delete.len()
         );
@@ -1325,6 +1422,40 @@ impl IndexScheduler {
         }
     }
 
+    // TODO: consider using a type alias or a struct embedder/template
+    pub fn embedders(
+        &self,
+        embedding_configs: Vec<(String, milli::vector::EmbeddingConfig)>,
+    ) -> Result<EmbeddingConfigs> {
+        let res: Result<_> = embedding_configs
+            .into_iter()
+            .map(|(name, milli::vector::EmbeddingConfig { embedder_options, prompt })| {
+                let prompt =
+                    Arc::new(prompt.try_into().map_err(meilisearch_types::milli::Error::from)?);
+                // optimistically return existing embedder
+                {
+                    let embedders = self.embedders.read().unwrap();
+                    if let Some(embedder) = embedders.get(&embedder_options) {
+                        return Ok((name, (embedder.clone(), prompt)));
+                    }
+                }
+
+                // add missing embedder
+                let embedder = Arc::new(
+                    Embedder::new(embedder_options.clone())
+                        .map_err(meilisearch_types::milli::vector::Error::from)
+                        .map_err(meilisearch_types::milli::Error::from)?,
+                );
+                {
+                    let mut embedders = self.embedders.write().unwrap();
+                    embedders.insert(embedder_options, embedder.clone());
+                }
+                Ok((name, (embedder, prompt)))
+            })
+            .collect();
+        res.map(EmbeddingConfigs::new)
+    }
+
     /// Blocks the thread until the test handle asks to progress to/through this breakpoint.
     ///
     /// Two messages are sent through the channel for each breakpoint.
@@ -1630,6 +1761,8 @@ mod tests {
                 indexes_path: tempdir.path().join("indexes"),
                 snapshots_path: tempdir.path().join("snapshots"),
                 dumps_path: tempdir.path().join("dumps"),
+                webhook_url: None,
+                webhook_authorization_header: None,
                 task_db_size: 1000 * 1000, // 1 MB, we don't use MiB on purpose.
                 index_base_map_size: 1000 * 1000, // 1 MB, we don't use MiB on purpose.
                 enable_mdb_writemap: false,
@@ -1638,6 +1771,7 @@ mod tests {
                 indexer_config,
                 autobatching_enabled: true,
                 max_number_of_tasks: 1_000_000,
+                max_number_of_batched_tasks: usize::MAX,
                 instance_features: Default::default(),
             };
             configuration(&mut options);
@@ -2110,10 +2244,7 @@ mod tests {
                 .unwrap();
             index_scheduler.assert_internally_consistent();
         }
-        for _ in 0..2 {
-            handle.advance_one_successful_batch();
-            index_scheduler.assert_internally_consistent();
-        }
+        handle.advance_one_successful_batch();
 
         snapshot!(snapshot_index_scheduler(&index_scheduler), name: "task_deletion_processed");
     }

index-scheduler/src/snapshots/lib.rs/task_deletion_delete_same_task_twice/task_deletion_processed.snap

@@ -34,12 +34,10 @@ catto: { number_of_documents: 1, field_distribution: {"id": 1} }
 [timestamp] [3,]
 ----------------------------------------------------------------------
 ### Started At:
-[timestamp] [2,]
-[timestamp] [3,]
+[timestamp] [2,3,]
 ----------------------------------------------------------------------
 ### Finished At:
-[timestamp] [2,]
-[timestamp] [3,]
+[timestamp] [2,3,]
 ----------------------------------------------------------------------
 ### File Store:
 00000000-0000-0000-0000-000000000001

meili-snap/Cargo.toml

@@ -11,6 +11,6 @@ edition.workspace = true
 license.workspace = true
 
 [dependencies]
-insta = { version = "^1.29.0", features = ["json", "redactions"] }
+insta = { version = "^1.34.0", features = ["json", "redactions"] }
 md5 = "0.7.0"
-once_cell = "1.17"
+once_cell = "1.19"

meilisearch-auth/Cargo.toml

@@ -11,16 +11,16 @@ edition.workspace = true
 license.workspace = true
 
 [dependencies]
-base64 = "0.21.0"
-enum-iterator = "1.4.0"
+base64 = "0.21.7"
+enum-iterator = "1.5.0"
 hmac = "0.12.1"
 maplit = "1.0.2"
 meilisearch-types = { path = "../meilisearch-types" }
 rand = "0.8.5"
-roaring = { version = "0.10.1", features = ["serde"] }
-serde = { version = "1.0.160", features = ["derive"] }
-serde_json = { version = "1.0.95", features = ["preserve_order"] }
-sha2 = "0.10.6"
-thiserror = "1.0.40"
-time = { version = "0.3.20", features = ["serde-well-known", "formatting", "parsing", "macros"] }
-uuid = { version = "1.3.1", features = ["serde", "v4"] }
+roaring = { version = "0.10.2", features = ["serde"] }
+serde = { version = "1.0.195", features = ["derive"] }
+serde_json = { version = "1.0.111", features = ["preserve_order"] }
+sha2 = "0.10.8"
+thiserror = "1.0.56"
+time = { version = "0.3.31", features = ["serde-well-known", "formatting", "parsing", "macros"] }
+uuid = { version = "1.6.1", features = ["serde", "v4"] }

meilisearch-types/Cargo.toml

@@ -11,31 +11,31 @@ edition.workspace = true
 license.workspace = true
 
 [dependencies]
-actix-web = { version = "4.3.1", default-features = false }
-anyhow = "1.0.70"
+actix-web = { version = "4.5.1", default-features = false }
+anyhow = "1.0.79"
 convert_case = "0.6.0"
-csv = "1.2.1"
-deserr = { version = "0.6.0", features = ["actix-web"] }
-either = { version = "1.8.1", features = ["serde"] }
-enum-iterator = "1.4.0"
+csv = "1.3.0"
+deserr = { version = "0.6.1", features = ["actix-web"] }
+either = { version = "1.9.0", features = ["serde"] }
+enum-iterator = "1.5.0"
 file-store = { path = "../file-store" }
-flate2 = "1.0.25"
+flate2 = "1.0.28"
 fst = "0.4.7"
 memmap2 = "0.7.1"
 milli = { path = "../milli" }
-roaring = { version = "0.10.1", features = ["serde"] }
-serde = { version = "1.0.160", features = ["derive"] }
+roaring = { version = "0.10.2", features = ["serde"] }
+serde = { version = "1.0.195", features = ["derive"] }
 serde-cs = "0.2.4"
-serde_json = "1.0.95"
-tar = "0.4.38"
-tempfile = "3.5.0"
-thiserror = "1.0.40"
-time = { version = "0.3.20", features = ["serde-well-known", "formatting", "parsing", "macros"] }
-tokio = "1.27"
-uuid = { version = "1.3.1", features = ["serde", "v4"] }
+serde_json = "1.0.111"
+tar = "0.4.40"
+tempfile = "3.9.0"
+thiserror = "1.0.56"
+time = { version = "0.3.31", features = ["serde-well-known", "formatting", "parsing", "macros"] }
+tokio = "1.35"
+uuid = { version = "1.6.1", features = ["serde", "v4"] }
 
 [dev-dependencies]
-insta = "1.29.0"
+insta = "1.34.0"
 meili-snap = { path = "../meili-snap" }
 
 [features]

meilisearch-types/src/deserr/mod.rs

@@ -188,3 +188,4 @@ merge_with_error_impl_take_error_message!(ParseOffsetDateTimeError);
 merge_with_error_impl_take_error_message!(ParseTaskKindError);
 merge_with_error_impl_take_error_message!(ParseTaskStatusError);
 merge_with_error_impl_take_error_message!(IndexUidFormatError);
+merge_with_error_impl_take_error_message!(InvalidSearchSemanticRatio);

meilisearch-types/src/error.rs

@@ -222,6 +222,8 @@ InvalidVectorsType                    , InvalidRequest       , BAD_REQUEST ;
 InvalidDocumentId                     , InvalidRequest       , BAD_REQUEST ;
 InvalidDocumentLimit                  , InvalidRequest       , BAD_REQUEST ;
 InvalidDocumentOffset                 , InvalidRequest       , BAD_REQUEST ;
+InvalidEmbedder                       , InvalidRequest       , BAD_REQUEST ;
+InvalidHybridQuery                    , InvalidRequest       , BAD_REQUEST ;
 InvalidIndexLimit                     , InvalidRequest       , BAD_REQUEST ;
 InvalidIndexOffset                    , InvalidRequest       , BAD_REQUEST ;
 InvalidIndexPrimaryKey                , InvalidRequest       , BAD_REQUEST ;
@@ -233,6 +235,7 @@ InvalidSearchAttributesToRetrieve     , InvalidRequest       , BAD_REQUEST ;
 InvalidSearchCropLength               , InvalidRequest       , BAD_REQUEST ;
 InvalidSearchCropMarker               , InvalidRequest       , BAD_REQUEST ;
 InvalidSearchFacets                   , InvalidRequest       , BAD_REQUEST ;
+InvalidSearchSemanticRatio            , InvalidRequest       , BAD_REQUEST ;
 InvalidFacetSearchFacetName           , InvalidRequest       , BAD_REQUEST ;
 InvalidSearchFilter                   , InvalidRequest       , BAD_REQUEST ;
 InvalidSearchHighlightPostTag         , InvalidRequest       , BAD_REQUEST ;
@@ -256,6 +259,7 @@ InvalidSettingsProximityPrecision     , InvalidRequest       , BAD_REQUEST ;
 InvalidSettingsFaceting               , InvalidRequest       , BAD_REQUEST ;
 InvalidSettingsFilterableAttributes   , InvalidRequest       , BAD_REQUEST ;
 InvalidSettingsPagination             , InvalidRequest       , BAD_REQUEST ;
+InvalidSettingsEmbedders              , InvalidRequest       , BAD_REQUEST ;
 InvalidSettingsRankingRules           , InvalidRequest       , BAD_REQUEST ;
 InvalidSettingsSearchableAttributes   , InvalidRequest       , BAD_REQUEST ;
 InvalidSettingsSortableAttributes     , InvalidRequest       , BAD_REQUEST ;
@@ -295,15 +299,20 @@ MissingFacetSearchFacetName           , InvalidRequest       , BAD_REQUEST ;
 MissingIndexUid                       , InvalidRequest       , BAD_REQUEST ;
 MissingMasterKey                      , Auth                 , UNAUTHORIZED ;
 MissingPayload                        , InvalidRequest       , BAD_REQUEST ;
+MissingSearchHybrid                   , InvalidRequest       , BAD_REQUEST ;
 MissingSwapIndexes                    , InvalidRequest       , BAD_REQUEST ;
 MissingTaskFilters                    , InvalidRequest       , BAD_REQUEST ;
 NoSpaceLeftOnDevice                   , System               , UNPROCESSABLE_ENTITY;
 PayloadTooLarge                       , InvalidRequest       , PAYLOAD_TOO_LARGE ;
 TaskNotFound                          , InvalidRequest       , NOT_FOUND ;
 TooManyOpenFiles                      , System               , UNPROCESSABLE_ENTITY ;
+TooManyVectors                        , InvalidRequest       , BAD_REQUEST ;
 UnretrievableDocument                 , Internal             , BAD_REQUEST ;
 UnretrievableErrorCode                , InvalidRequest       , BAD_REQUEST ;
-UnsupportedMediaType                  , InvalidRequest       , UNSUPPORTED_MEDIA_TYPE
+UnsupportedMediaType                  , InvalidRequest       , UNSUPPORTED_MEDIA_TYPE ;
+
+// Experimental features
+VectorEmbeddingError                  , InvalidRequest       , BAD_REQUEST
 }
 
 impl ErrorCode for JoinError {
@@ -336,6 +345,16 @@ impl ErrorCode for milli::Error {
                     UserError::InvalidDocumentId { .. } | UserError::TooManyDocumentIds { .. } => {
                         Code::InvalidDocumentId
                     }
+                    UserError::MissingDocumentField(_) => Code::InvalidDocumentFields,
+                    UserError::InvalidFieldForSource { .. }
+                    | UserError::MissingFieldForSource { .. }
+                    | UserError::InvalidOpenAiModel { .. }
+                    | UserError::InvalidOpenAiModelDimensions { .. }
+                    | UserError::InvalidOpenAiModelDimensionsMax { .. }
+                    | UserError::InvalidSettingsDimensions { .. }
+                    | UserError::InvalidPrompt(_) => Code::InvalidSettingsEmbedders,
+                    UserError::TooManyEmbedders(_) => Code::InvalidSettingsEmbedders,
+                    UserError::InvalidPromptForEmbeddings(..) => Code::InvalidSettingsEmbedders,
                     UserError::NoPrimaryKeyCandidateFound => Code::IndexPrimaryKeyNoCandidateFound,
                     UserError::MultiplePrimaryKeyCandidatesFound { .. } => {
                         Code::IndexPrimaryKeyMultipleCandidatesFound
@@ -353,11 +372,15 @@ impl ErrorCode for milli::Error {
                     UserError::CriterionError(_) => Code::InvalidSettingsRankingRules,
                     UserError::InvalidGeoField { .. } => Code::InvalidDocumentGeoField,
                     UserError::InvalidVectorDimensions { .. } => Code::InvalidVectorDimensions,
+                    UserError::InvalidVectorsMapType { .. } => Code::InvalidVectorsType,
                     UserError::InvalidVectorsType { .. } => Code::InvalidVectorsType,
+                    UserError::TooManyVectors(_, _) => Code::TooManyVectors,
                     UserError::SortError(_) => Code::InvalidSearchSort,
                     UserError::InvalidMinTypoWordLenSetting(_, _) => {
                         Code::InvalidSettingsTypoTolerance
                     }
+                    UserError::InvalidEmbedder(_) => Code::InvalidEmbedder,
+                    UserError::VectorEmbeddingError(_) => Code::VectorEmbeddingError,
                 }
             }
         }
@@ -445,6 +468,15 @@ impl fmt::Display for DeserrParseIntError {
     }
 }
 
+impl fmt::Display for deserr_codes::InvalidSearchSemanticRatio {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        write!(
+            f,
+            "the value of `semanticRatio` is invalid, expected a float between `0.0` and `1.0`."
+        )
+    }
+}
+
 #[macro_export]
 macro_rules! internal_error {
     ($target:ty : $($other:path), *) => {

meilisearch-types/src/features.rs

@@ -3,14 +3,14 @@ use serde::{Deserialize, Serialize};
 #[derive(Serialize, Deserialize, Debug, Clone, Copy, Default, PartialEq, Eq)]
 #[serde(rename_all = "camelCase", default)]
 pub struct RuntimeTogglableFeatures {
-    pub score_details: bool,
     pub vector_store: bool,
     pub metrics: bool,
+    pub logs_route: bool,
     pub export_puffin_reports: bool,
-    pub proximity_precision: bool,
 }
 
 #[derive(Default, Debug, Clone, Copy)]
 pub struct InstanceTogglableFeatures {
     pub metrics: bool,
+    pub logs_route: bool,
 }

meilisearch-types/src/lib.rs

@@ -9,6 +9,7 @@ pub mod index_uid_pattern;
 pub mod keys;
 pub mod settings;
 pub mod star_or;
+pub mod task_view;
 pub mod tasks;
 pub mod versioning;
 pub use milli::{heed, Index};

meilisearch-types/src/settings.rs

@@ -199,6 +199,10 @@ pub struct Settings<T> {
     #[deserr(default, error = DeserrJsonError<InvalidSettingsPagination>)]
     pub pagination: Setting<PaginationSettings>,
 
+    #[serde(default, skip_serializing_if = "Setting::is_not_set")]
+    #[deserr(default, error = DeserrJsonError<InvalidSettingsEmbedders>)]
+    pub embedders: Setting<BTreeMap<String, Setting<milli::vector::settings::EmbeddingSettings>>>,
+
     #[serde(skip)]
     #[deserr(skip)]
     pub _kind: PhantomData<T>,
@@ -222,6 +226,7 @@ impl Settings<Checked> {
             typo_tolerance: Setting::Reset,
             faceting: Setting::Reset,
             pagination: Setting::Reset,
+            embedders: Setting::Reset,
             _kind: PhantomData,
         }
     }
@@ -243,6 +248,7 @@ impl Settings<Checked> {
             typo_tolerance,
             faceting,
             pagination,
+            embedders,
             ..
         } = self;
 
@@ -262,6 +268,7 @@ impl Settings<Checked> {
             typo_tolerance,
             faceting,
             pagination,
+            embedders,
             _kind: PhantomData,
         }
     }
@@ -307,9 +314,25 @@ impl Settings<Unchecked> {
             typo_tolerance: self.typo_tolerance,
             faceting: self.faceting,
             pagination: self.pagination,
+            embedders: self.embedders,
             _kind: PhantomData,
         }
     }
+
+    pub fn validate(self) -> Result<Self, milli::Error> {
+        self.validate_embedding_settings()
+    }
+
+    fn validate_embedding_settings(mut self) -> Result<Self, milli::Error> {
+        let Setting::Set(mut configs) = self.embedders else { return Ok(self) };
+        for (name, config) in configs.iter_mut() {
+            let config_to_check = std::mem::take(config);
+            let checked_config = milli::update::validate_embedding_settings(config_to_check, name)?;
+            *config = checked_config
+        }
+        self.embedders = Setting::Set(configs);
+        Ok(self)
+    }
 }
 
 #[derive(Debug, Clone, Serialize, Deserialize)]
@@ -490,6 +513,12 @@ pub fn apply_settings_to_builder(
         Setting::Reset => builder.reset_pagination_max_total_hits(),
         Setting::NotSet => (),
     }
+
+    match settings.embedders.clone() {
+        Setting::Set(value) => builder.set_embedder_settings(value),
+        Setting::Reset => builder.reset_embedder_settings(),
+        Setting::NotSet => (),
+    }
 }
 
 pub fn settings(
@@ -571,6 +600,13 @@ pub fn settings(
         ),
     };
 
+    let embedders: BTreeMap<_, _> = index
+        .embedding_configs(rtxn)?
+        .into_iter()
+        .map(|(name, config)| (name, Setting::Set(config.into())))
+        .collect();
+    let embedders = if embedders.is_empty() { Setting::NotSet } else { Setting::Set(embedders) };
+
     Ok(Settings {
         displayed_attributes: match displayed_attributes {
             Some(attrs) => Setting::Set(attrs),
@@ -591,14 +627,12 @@ pub fn settings(
             Some(field) => Setting::Set(field),
             None => Setting::Reset,
         },
-        proximity_precision: match proximity_precision {
-            Some(precision) => Setting::Set(precision),
-            None => Setting::Reset,
-        },
+        proximity_precision: Setting::Set(proximity_precision.unwrap_or_default()),
         synonyms: Setting::Set(synonyms),
         typo_tolerance: Setting::Set(typo_tolerance),
         faceting: Setting::Set(faceting),
         pagination: Setting::Set(pagination),
+        embedders,
         _kind: PhantomData,
     })
 }
@@ -699,27 +733,28 @@ impl From<RankingRuleView> for Criterion {
     }
 }
 
-#[derive(Debug, Clone, Copy, PartialEq, Eq, Deserr, Serialize, Deserialize)]
+#[derive(Default, Debug, Clone, Copy, PartialEq, Eq, Deserr, Serialize, Deserialize)]
 #[serde(deny_unknown_fields, rename_all = "camelCase")]
 #[deserr(error = DeserrJsonError<InvalidSettingsProximityPrecision>, rename_all = camelCase, deny_unknown_fields)]
 pub enum ProximityPrecisionView {
-    WordScale,
-    AttributeScale,
+    #[default]
+    ByWord,
+    ByAttribute,
 }
 
 impl From<ProximityPrecision> for ProximityPrecisionView {
     fn from(value: ProximityPrecision) -> Self {
         match value {
-            ProximityPrecision::WordScale => ProximityPrecisionView::WordScale,
-            ProximityPrecision::AttributeScale => ProximityPrecisionView::AttributeScale,
+            ProximityPrecision::ByWord => ProximityPrecisionView::ByWord,
+            ProximityPrecision::ByAttribute => ProximityPrecisionView::ByAttribute,
         }
     }
 }
 impl From<ProximityPrecisionView> for ProximityPrecision {
     fn from(value: ProximityPrecisionView) -> Self {
         match value {
-            ProximityPrecisionView::WordScale => ProximityPrecision::WordScale,
-            ProximityPrecisionView::AttributeScale => ProximityPrecision::AttributeScale,
+            ProximityPrecisionView::ByWord => ProximityPrecision::ByWord,
+            ProximityPrecisionView::ByAttribute => ProximityPrecision::ByAttribute,
         }
     }
 }
@@ -747,6 +782,7 @@ pub(crate) mod test {
             typo_tolerance: Setting::NotSet,
             faceting: Setting::NotSet,
             pagination: Setting::NotSet,
+            embedders: Setting::NotSet,
             _kind: PhantomData::<Unchecked>,
         };
 
@@ -772,6 +808,7 @@ pub(crate) mod test {
             typo_tolerance: Setting::NotSet,
             faceting: Setting::NotSet,
             pagination: Setting::NotSet,
+            embedders: Setting::NotSet,
             _kind: PhantomData::<Unchecked>,
         };
 

meilisearch-types/src/task_view.rs

@@ -0,0 +1,139 @@
+use serde::Serialize;
+use time::{Duration, OffsetDateTime};
+
+use crate::error::ResponseError;
+use crate::settings::{Settings, Unchecked};
+use crate::tasks::{serialize_duration, Details, IndexSwap, Kind, Status, Task, TaskId};
+
+#[derive(Debug, Clone, PartialEq, Eq, Serialize)]
+#[serde(rename_all = "camelCase")]
+pub struct TaskView {
+    pub uid: TaskId,
+    #[serde(default)]
+    pub index_uid: Option<String>,
+    pub status: Status,
+    #[serde(rename = "type")]
+    pub kind: Kind,
+    pub canceled_by: Option<TaskId>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub details: Option<DetailsView>,
+    pub error: Option<ResponseError>,
+    #[serde(serialize_with = "serialize_duration", default)]
+    pub duration: Option<Duration>,
+    #[serde(with = "time::serde::rfc3339")]
+    pub enqueued_at: OffsetDateTime,
+    #[serde(with = "time::serde::rfc3339::option", default)]
+    pub started_at: Option<OffsetDateTime>,
+    #[serde(with = "time::serde::rfc3339::option", default)]
+    pub finished_at: Option<OffsetDateTime>,
+}
+
+impl TaskView {
+    pub fn from_task(task: &Task) -> TaskView {
+        TaskView {
+            uid: task.uid,
+            index_uid: task.index_uid().map(ToOwned::to_owned),
+            status: task.status,
+            kind: task.kind.as_kind(),
+            canceled_by: task.canceled_by,
+            details: task.details.clone().map(DetailsView::from),
+            error: task.error.clone(),
+            duration: task.started_at.zip(task.finished_at).map(|(start, end)| end - start),
+            enqueued_at: task.enqueued_at,
+            started_at: task.started_at,
+            finished_at: task.finished_at,
+        }
+    }
+}
+
+#[derive(Default, Debug, PartialEq, Eq, Clone, Serialize)]
+#[serde(rename_all = "camelCase")]
+pub struct DetailsView {
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub received_documents: Option<u64>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub indexed_documents: Option<Option<u64>>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub primary_key: Option<Option<String>>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub provided_ids: Option<usize>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub deleted_documents: Option<Option<u64>>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub matched_tasks: Option<u64>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub canceled_tasks: Option<Option<u64>>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub deleted_tasks: Option<Option<u64>>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub original_filter: Option<Option<String>>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub dump_uid: Option<Option<String>>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    #[serde(flatten)]
+    pub settings: Option<Box<Settings<Unchecked>>>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub swaps: Option<Vec<IndexSwap>>,
+}
+
+impl From<Details> for DetailsView {
+    fn from(details: Details) -> Self {
+        match details {
+            Details::DocumentAdditionOrUpdate { received_documents, indexed_documents } => {
+                DetailsView {
+                    received_documents: Some(received_documents),
+                    indexed_documents: Some(indexed_documents),
+                    ..DetailsView::default()
+                }
+            }
+            Details::SettingsUpdate { settings } => {
+                DetailsView { settings: Some(settings), ..DetailsView::default() }
+            }
+            Details::IndexInfo { primary_key } => {
+                DetailsView { primary_key: Some(primary_key), ..DetailsView::default() }
+            }
+            Details::DocumentDeletion {
+                provided_ids: received_document_ids,
+                deleted_documents,
+            } => DetailsView {
+                provided_ids: Some(received_document_ids),
+                deleted_documents: Some(deleted_documents),
+                original_filter: Some(None),
+                ..DetailsView::default()
+            },
+            Details::DocumentDeletionByFilter { original_filter, deleted_documents } => {
+                DetailsView {
+                    provided_ids: Some(0),
+                    original_filter: Some(Some(original_filter)),
+                    deleted_documents: Some(deleted_documents),
+                    ..DetailsView::default()
+                }
+            }
+            Details::ClearAll { deleted_documents } => {
+                DetailsView { deleted_documents: Some(deleted_documents), ..DetailsView::default() }
+            }
+            Details::TaskCancelation { matched_tasks, canceled_tasks, original_filter } => {
+                DetailsView {
+                    matched_tasks: Some(matched_tasks),
+                    canceled_tasks: Some(canceled_tasks),
+                    original_filter: Some(Some(original_filter)),
+                    ..DetailsView::default()
+                }
+            }
+            Details::TaskDeletion { matched_tasks, deleted_tasks, original_filter } => {
+                DetailsView {
+                    matched_tasks: Some(matched_tasks),
+                    deleted_tasks: Some(deleted_tasks),
+                    original_filter: Some(Some(original_filter)),
+                    ..DetailsView::default()
+                }
+            }
+            Details::Dump { dump_uid } => {
+                DetailsView { dump_uid: Some(dump_uid), ..DetailsView::default() }
+            }
+            Details::IndexSwap { swaps } => {
+                DetailsView { swaps: Some(swaps), ..Default::default() }
+            }
+        }
+    }
+}

meilisearch/Cargo.toml

@@ -13,123 +13,126 @@ license.workspace = true
 default-run = "meilisearch"
 
 [dependencies]
-actix-cors = "0.6.4"
-actix-http = { version = "3.3.1", default-features = false, features = [
+actix-cors = "0.7.0"
+actix-http = { version = "3.6.0", default-features = false, features = [
     "compress-brotli",
     "compress-gzip",
-    "rustls",
+    "rustls-0_21",
 ] }
 actix-utils = "3.0.1"
-actix-web = { version = "4.3.1", default-features = false, features = [
+actix-web = { version = "4.5.1", default-features = false, features = [
     "macros",
     "compress-brotli",
     "compress-gzip",
     "cookies",
-    "rustls",
+    "rustls-0_21",
 ] }
 actix-web-static-files = { git = "https://github.com/kilork/actix-web-static-files.git", rev = "2d3b6160", optional = true }
-anyhow = { version = "1.0.70", features = ["backtrace"] }
+anyhow = { version = "1.0.79", features = ["backtrace"] }
 async-stream = "0.3.5"
-async-trait = "0.1.68"
-bstr = "1.4.0"
+async-trait = "0.1.77"
+bstr = "1.9.0"
 byte-unit = { version = "4.0.19", default-features = false, features = [
     "std",
     "serde",
 ] }
-bytes = "1.4.0"
-clap = { version = "4.2.1", features = ["derive", "env"] }
-crossbeam-channel = "0.5.8"
-deserr = { version = "0.6.0", features = ["actix-web"] }
+bytes = "1.5.0"
+clap = { version = "4.4.17", features = ["derive", "env"] }
+crossbeam-channel = "0.5.11"
+deserr = { version = "0.6.1", features = ["actix-web"] }
 dump = { path = "../dump" }
-either = "1.8.1"
-env_logger = "0.10.0"
+either = "1.9.0"
 file-store = { path = "../file-store" }
-flate2 = "1.0.25"
+flate2 = "1.0.28"
 fst = "0.4.7"
-futures = "0.3.28"
-futures-util = "0.3.28"
-http = "0.2.9"
+futures = "0.3.30"
+futures-util = "0.3.30"
+http = "0.2.11"
 index-scheduler = { path = "../index-scheduler" }
-indexmap = { version = "2.0.0", features = ["serde"] }
-is-terminal = "0.4.8"
+indexmap = { version = "2.1.0", features = ["serde"] }
+is-terminal = "0.4.10"
 itertools = "0.11.0"
-jsonwebtoken = "8.3.0"
+jsonwebtoken = "9.2.0"
 lazy_static = "1.4.0"
-log = "0.4.17"
 meilisearch-auth = { path = "../meilisearch-auth" }
 meilisearch-types = { path = "../meilisearch-types" }
-mimalloc = { version = "0.1.37", default-features = false }
+mimalloc = { version = "0.1.39", default-features = false }
 mime = "0.3.17"
-num_cpus = "1.15.0"
-obkv = "0.2.0"
-once_cell = "1.17.1"
-ordered-float = "3.7.0"
+num_cpus = "1.16.0"
+obkv = "0.2.1"
+once_cell = "1.19.0"
+ordered-float = "4.2.0"
 parking_lot = "0.12.1"
 permissive-json-pointer = { path = "../permissive-json-pointer" }
-pin-project-lite = "0.2.9"
+pin-project-lite = "0.2.13"
 platform-dirs = "0.3.0"
 prometheus = { version = "0.13.3", features = ["process"] }
 puffin = { version = "0.16.0", features = ["serialization"] }
 rand = "0.8.5"
-rayon = "1.7.0"
-regex = "1.7.3"
-reqwest = { version = "0.11.16", features = [
+rayon = "1.8.0"
+regex = "1.10.2"
+reqwest = { version = "0.11.23", features = [
     "rustls-tls",
     "json",
 ], default-features = false }
-rustls = "0.20.8"
+rustls = "0.21.6"
 rustls-pemfile = "1.0.2"
-segment = { version = "0.2.2", optional = true }
-serde = { version = "1.0.160", features = ["derive"] }
-serde_json = { version = "1.0.95", features = ["preserve_order"] }
-sha2 = "0.10.6"
-siphasher = "0.3.10"
-slice-group-by = "0.3.0"
+segment = { version = "0.2.3", optional = true }
+serde = { version = "1.0.195", features = ["derive"] }
+serde_json = { version = "1.0.111", features = ["preserve_order"] }
+sha2 = "0.10.8"
+siphasher = "1.0.0"
+slice-group-by = "0.3.1"
 static-files = { version = "0.2.3", optional = true }
-sysinfo = "0.29.7"
-tar = "0.4.38"
-tempfile = "3.5.0"
-thiserror = "1.0.40"
-time = { version = "0.3.20", features = [
+sysinfo = "0.30.5"
+tar = "0.4.40"
+tempfile = "3.9.0"
+thiserror = "1.0.56"
+time = { version = "0.3.31", features = [
     "serde-well-known",
     "formatting",
     "parsing",
     "macros",
 ] }
-tokio = { version = "1.27.0", features = ["full"] }
-tokio-stream = "0.1.12"
-toml = "0.7.3"
-uuid = { version = "1.3.1", features = ["serde", "v4"] }
-walkdir = "2.3.3"
+tokio = { version = "1.35.1", features = ["full"] }
+tokio-stream = "0.1.14"
+toml = "0.8.8"
+uuid = { version = "1.6.1", features = ["serde", "v4"] }
+walkdir = "2.4.0"
 yaup = "0.2.1"
 serde_urlencoded = "0.7.1"
-termcolor = "1.2.0"
+termcolor = "1.4.1"
+url = { version = "2.5.0", features = ["serde"] }
+tracing = "0.1.40"
+tracing-subscriber = "0.3.18"
+tracing-trace = { version = "0.1.0", path = "../tracing-trace" }
+tracing-actix-web = "0.7.9"
 
 [dev-dependencies]
-actix-rt = "2.8.0"
+actix-rt = "2.9.0"
 assert-json-diff = "2.0.2"
-brotli = "3.3.4"
-insta = "1.29.0"
-manifest-dir-macros = "0.1.16"
+brotli = "3.4.0"
+insta = "1.34.0"
+manifest-dir-macros = "0.1.18"
 maplit = "1.0.2"
 meili-snap = { path = "../meili-snap" }
-temp-env = "0.3.3"
-urlencoding = "2.1.2"
+temp-env = "0.3.6"
+urlencoding = "2.1.3"
 yaup = "0.2.1"
 
 [build-dependencies]
-anyhow = { version = "1.0.70", optional = true }
-cargo_toml = { version = "0.15.2", optional = true }
+anyhow = { version = "1.0.79", optional = true }
+cargo_toml = { version = "0.18.0", optional = true }
 hex = { version = "0.4.3", optional = true }
-reqwest = { version = "0.11.16", features = [
+reqwest = { version = "0.11.23", features = [
     "blocking",
     "rustls-tls",
 ], default-features = false, optional = true }
 sha-1 = { version = "0.10.1", optional = true }
 static-files = { version = "0.2.3", optional = true }
-tempfile = { version = "3.5.0", optional = true }
+tempfile = { version = "3.9.0", optional = true }
 vergen = { version = "7.5.1", default-features = false, features = ["git"] }
-zip = { version = "0.6.4", optional = true }
+zip = { version = "0.6.6", optional = true }
 
 [features]
 default = ["analytics", "meilisearch-types/all-tokenizations", "mini-dashboard"]
@@ -153,5 +156,5 @@ greek = ["meilisearch-types/greek"]
 khmer = ["meilisearch-types/khmer"]
 
 [package.metadata.mini-dashboard]
-assets-url = "https://github.com/meilisearch/mini-dashboard/releases/download/v0.2.11/build.zip"
-sha1 = "83cd44ed1e5f97ecb581dc9f958a63f4ccc982d9"
+assets-url = "https://github.com/meilisearch/mini-dashboard/releases/download/v0.2.13/build.zip"
+sha1 = "e20cc9b390003c6c844f4b8bcc5c5013191a77ff"

meilisearch/src/analytics/segment_analytics.rs

@@ -18,7 +18,7 @@ use segment::message::{Identify, Track, User};
 use segment::{AutoBatcher, Batcher, HttpClient};
 use serde::Serialize;
 use serde_json::{json, Value};
-use sysinfo::{DiskExt, System, SystemExt};
+use sysinfo::{Disks, System};
 use time::OffsetDateTime;
 use tokio::select;
 use tokio::sync::mpsc::{self, Receiver, Sender};
@@ -36,7 +36,7 @@ use crate::routes::{create_all_stats, Stats};
 use crate::search::{
     FacetSearchResult, MatchingStrategy, SearchQuery, SearchQueryWithIndex, SearchResult,
     DEFAULT_CROP_LENGTH, DEFAULT_CROP_MARKER, DEFAULT_HIGHLIGHT_POST_TAG,
-    DEFAULT_HIGHLIGHT_PRE_TAG, DEFAULT_SEARCH_LIMIT,
+    DEFAULT_HIGHLIGHT_PRE_TAG, DEFAULT_SEARCH_LIMIT, DEFAULT_SEMANTIC_RATIO,
 };
 use crate::Opt;
 
@@ -250,7 +250,9 @@ impl super::Analytics for SegmentAnalytics {
 struct Infos {
     env: String,
     experimental_enable_metrics: bool,
+    experimental_enable_logs_route: bool,
     experimental_reduce_indexing_memory_usage: bool,
+    experimental_max_number_of_batched_tasks: usize,
     db_path: bool,
     import_dump: bool,
     dump_dir: bool,
@@ -263,6 +265,8 @@ struct Infos {
     ignore_snapshot_if_db_exists: bool,
     http_addr: bool,
     http_payload_size_limit: Byte,
+    task_queue_webhook: bool,
+    task_webhook_authorization_header: bool,
     log_level: String,
     max_indexing_memory: MaxMemory,
     max_indexing_threads: MaxThreads,
@@ -284,10 +288,14 @@ impl From<Opt> for Infos {
         let Opt {
             db_path,
             experimental_enable_metrics,
+            experimental_enable_logs_route,
             experimental_reduce_indexing_memory_usage,
+            experimental_max_number_of_batched_tasks,
             http_addr,
             master_key: _,
             env,
+            task_webhook_url,
+            task_webhook_authorization_header,
             max_index_size: _,
             max_task_db_size: _,
             http_payload_size_limit,
@@ -327,6 +335,7 @@ impl From<Opt> for Infos {
         Self {
             env,
             experimental_enable_metrics,
+            experimental_enable_logs_route,
             experimental_reduce_indexing_memory_usage,
             db_path: db_path != PathBuf::from("./data.ms"),
             import_dump: import_dump.is_some(),
@@ -340,6 +349,9 @@ impl From<Opt> for Infos {
             ignore_snapshot_if_db_exists,
             http_addr: http_addr != default_http_addr(),
             http_payload_size_limit,
+            experimental_max_number_of_batched_tasks,
+            task_queue_webhook: task_webhook_url.is_some(),
+            task_webhook_authorization_header: task_webhook_authorization_header.is_some(),
             log_level: log_level.to_string(),
             max_indexing_memory,
             max_indexing_threads,
@@ -377,16 +389,17 @@ impl Segment {
     fn compute_traits(opt: &Opt, stats: Stats) -> Value {
         static FIRST_START_TIMESTAMP: Lazy<Instant> = Lazy::new(Instant::now);
         static SYSTEM: Lazy<Value> = Lazy::new(|| {
+            let disks = Disks::new_with_refreshed_list();
             let mut sys = System::new_all();
             sys.refresh_all();
-            let kernel_version =
-                sys.kernel_version().and_then(|k| k.split_once('-').map(|(k, _)| k.to_string()));
+            let kernel_version = System::kernel_version()
+                .and_then(|k| k.split_once('-').map(|(k, _)| k.to_string()));
             json!({
-                    "distribution": sys.name(),
+                    "distribution": System::name(),
                     "kernel_version": kernel_version,
                     "cores": sys.cpus().len(),
                     "ram_size": sys.total_memory(),
-                    "disk_size": sys.disks().iter().map(|disk| disk.total_space()).max(),
+                    "disk_size": disks.iter().map(|disk| disk.total_space()).max(),
                     "server_provider": std::env::var("MEILI_SERVER_PROVIDER").ok(),
             })
         });
@@ -583,6 +596,11 @@ pub struct SearchAggregator {
     // vector
     // The maximum number of floats in a vector request
     max_vector_size: usize,
+    // Whether the semantic ratio passed to a hybrid search equals the default ratio.
+    semantic_ratio: bool,
+    // Whether a non-default embedder was specified
+    embedder: bool,
+    hybrid: bool,
 
     // every time a search is done, we increment the counter linked to the used settings
     matching_strategy: HashMap<String, usize>,
@@ -636,6 +654,7 @@ impl SearchAggregator {
             crop_marker,
             matching_strategy,
             attributes_to_search_on,
+            hybrid,
         } = query;
 
         let mut ret = Self::default();
@@ -709,6 +728,12 @@ impl SearchAggregator {
         ret.show_ranking_score = *show_ranking_score;
         ret.show_ranking_score_details = *show_ranking_score_details;
 
+        if let Some(hybrid) = hybrid {
+            ret.semantic_ratio = hybrid.semantic_ratio != DEFAULT_SEMANTIC_RATIO();
+            ret.embedder = hybrid.embedder.is_some();
+            ret.hybrid = true;
+        }
+
         ret
     }
 
@@ -762,6 +787,9 @@ impl SearchAggregator {
             facets_total_number_of_facets,
             show_ranking_score,
             show_ranking_score_details,
+            semantic_ratio,
+            embedder,
+            hybrid,
         } = other;
 
         if self.timestamp.is_none() {
@@ -807,6 +835,9 @@ impl SearchAggregator {
 
         // vector
         self.max_vector_size = self.max_vector_size.max(max_vector_size);
+        self.semantic_ratio |= semantic_ratio;
+        self.hybrid |= hybrid;
+        self.embedder |= embedder;
 
         // pagination
         self.max_limit = self.max_limit.max(max_limit);
@@ -875,6 +906,9 @@ impl SearchAggregator {
             facets_total_number_of_facets,
             show_ranking_score,
             show_ranking_score_details,
+            semantic_ratio,
+            embedder,
+            hybrid,
         } = self;
 
         if total_received == 0 {
@@ -914,6 +948,11 @@ impl SearchAggregator {
                 "vector": {
                     "max_vector_size": max_vector_size,
                 },
+                "hybrid": {
+                    "enabled": hybrid,
+                    "semantic_ratio": semantic_ratio,
+                    "embedder": embedder,
+                },
                 "pagination": {
                    "max_limit": max_limit,
                    "max_offset": max_offset,
@@ -1009,6 +1048,7 @@ impl MultiSearchAggregator {
                     crop_marker: _,
                     matching_strategy: _,
                     attributes_to_search_on: _,
+                    hybrid: _,
                 } = query;
 
                 index_uid.as_str()
@@ -1155,6 +1195,7 @@ impl FacetSearchAggregator {
             filter,
             matching_strategy,
             attributes_to_search_on,
+            hybrid,
         } = query;
 
         let mut ret = Self::default();
@@ -1168,7 +1209,8 @@ impl FacetSearchAggregator {
             || vector.is_some()
             || filter.is_some()
             || *matching_strategy != MatchingStrategy::default()
-            || attributes_to_search_on.is_some();
+            || attributes_to_search_on.is_some()
+            || hybrid.is_some();
 
         ret
     }

meilisearch/src/error.rs

@@ -12,6 +12,8 @@ pub enum MeilisearchHttpError {
     #[error("A Content-Type header is missing. Accepted values for the Content-Type header are: {}",
             .0.iter().map(|s| format!("`{}`", s)).collect::<Vec<_>>().join(", "))]
     MissingContentType(Vec<String>),
+    #[error("The `/logs/stream` route is currently in use by someone else.")]
+    AlreadyUsedLogRoute,
     #[error("The Content-Type `{0}` does not support the use of a csv delimiter. The csv delimiter can only be used with the Content-Type `text/csv`.")]
     CsvDelimiterWithWrongContentType(String),
     #[error(
@@ -51,12 +53,15 @@ pub enum MeilisearchHttpError {
     DocumentFormat(#[from] DocumentFormatError),
     #[error(transparent)]
     Join(#[from] JoinError),
+    #[error("Invalid request: missing `hybrid` parameter when both `q` and `vector` are present.")]
+    MissingSearchHybrid,
 }
 
 impl ErrorCode for MeilisearchHttpError {
     fn error_code(&self) -> Code {
         match self {
             MeilisearchHttpError::MissingContentType(_) => Code::MissingContentType,
+            MeilisearchHttpError::AlreadyUsedLogRoute => Code::BadRequest,
             MeilisearchHttpError::CsvDelimiterWithWrongContentType(_) => Code::InvalidContentType,
             MeilisearchHttpError::MissingPayload(_) => Code::MissingPayload,
             MeilisearchHttpError::InvalidContentType(_, _) => Code::InvalidContentType,
@@ -74,6 +79,7 @@ impl ErrorCode for MeilisearchHttpError {
             MeilisearchHttpError::FileStore(_) => Code::Internal,
             MeilisearchHttpError::DocumentFormat(e) => e.error_code(),
             MeilisearchHttpError::Join(_) => Code::Internal,
+            MeilisearchHttpError::MissingSearchHybrid => Code::MissingSearchHybrid,
         }
     }
 }

meilisearch/src/lib.rs

@@ -29,7 +29,6 @@ use error::PayloadError;
 use extractors::payload::PayloadConfig;
 use http::header::CONTENT_TYPE;
 use index_scheduler::{IndexScheduler, IndexSchedulerOptions};
-use log::error;
 use meilisearch_auth::AuthController;
 use meilisearch_types::milli::documents::{DocumentsBatchBuilder, DocumentsBatchReader};
 use meilisearch_types::milli::update::{IndexDocumentsConfig, IndexDocumentsMethod};
@@ -39,6 +38,8 @@ use meilisearch_types::versioning::{check_version_file, create_version_file};
 use meilisearch_types::{compression, milli, VERSION_FILE_NAME};
 pub use option::Opt;
 use option::ScheduleSnapshot;
+use tracing::{error, info_span};
+use tracing_subscriber::filter::Targets;
 
 use crate::error::MeilisearchHttpError;
 
@@ -86,10 +87,21 @@ fn is_empty_db(db_path: impl AsRef<Path>) -> bool {
     }
 }
 
+/// The handle used to update the logs at runtime. Must be accessible from the `main.rs` and the `route/logs.rs`.
+pub type LogRouteHandle =
+    tracing_subscriber::reload::Handle<LogRouteType, tracing_subscriber::Registry>;
+
+pub type LogRouteType = tracing_subscriber::filter::Filtered<
+    Option<Box<dyn tracing_subscriber::Layer<tracing_subscriber::Registry> + Send + Sync>>,
+    Targets,
+    tracing_subscriber::Registry,
+>;
+
 pub fn create_app(
     index_scheduler: Data<IndexScheduler>,
     auth_controller: Data<AuthController>,
     opt: Opt,
+    logs: LogRouteHandle,
     analytics: Arc<dyn Analytics>,
     enable_dashboard: bool,
 ) -> actix_web::App<
@@ -108,6 +120,7 @@ pub fn create_app(
                 index_scheduler.clone(),
                 auth_controller.clone(),
                 &opt,
+                logs,
                 analytics.clone(),
             )
         })
@@ -123,11 +136,49 @@ pub fn create_app(
             .allow_any_method()
             .max_age(86_400), // 24h
     )
-    .wrap(actix_web::middleware::Logger::default())
+    .wrap(tracing_actix_web::TracingLogger::<AwebTracingLogger>::new())
     .wrap(actix_web::middleware::Compress::default())
     .wrap(actix_web::middleware::NormalizePath::new(actix_web::middleware::TrailingSlash::Trim))
 }
 
+struct AwebTracingLogger;
+
+impl tracing_actix_web::RootSpanBuilder for AwebTracingLogger {
+    fn on_request_start(request: &actix_web::dev::ServiceRequest) -> tracing::Span {
+        use tracing::field::Empty;
+
+        let conn_info = request.connection_info();
+        let headers = request.headers();
+        let user_agent = headers
+            .get(http::header::USER_AGENT)
+            .map(|value| String::from_utf8_lossy(value.as_bytes()).into_owned())
+            .unwrap_or_default();
+        info_span!("HTTP request", method = %request.method(), host = conn_info.host(), route = %request.path(), query_parameters = %request.query_string(), %user_agent, status_code = Empty, error = Empty)
+    }
+
+    fn on_request_end<B: MessageBody>(
+        span: tracing::Span,
+        outcome: &Result<ServiceResponse<B>, actix_web::Error>,
+    ) {
+        match &outcome {
+            Ok(response) => {
+                let code: i32 = response.response().status().as_u16().into();
+                span.record("status_code", code);
+
+                if let Some(error) = response.response().error() {
+                    // use the status code already constructed for the outgoing HTTP response
+                    span.record("error", &tracing::field::display(error.as_response_error()));
+                }
+            }
+            Err(error) => {
+                let code: i32 = error.error_response().status().as_u16().into();
+                span.record("status_code", code);
+                span.record("error", &tracing::field::display(error.as_response_error()));
+            }
+        };
+    }
+}
+
 enum OnFailure {
     RemoveDb,
     KeepDb,
@@ -228,12 +279,15 @@ fn open_or_create_database_unchecked(
             indexes_path: opt.db_path.join("indexes"),
             snapshots_path: opt.snapshot_dir.clone(),
             dumps_path: opt.dump_dir.clone(),
+            webhook_url: opt.task_webhook_url.as_ref().map(|url| url.to_string()),
+            webhook_authorization_header: opt.task_webhook_authorization_header.clone(),
             task_db_size: opt.max_task_db_size.get_bytes() as usize,
             index_base_map_size: opt.max_index_size.get_bytes() as usize,
             enable_mdb_writemap: opt.experimental_reduce_indexing_memory_usage,
             indexer_config: (&opt.indexer_options).try_into()?,
             autobatching_enabled: true,
             max_number_of_tasks: 1_000_000,
+            max_number_of_batched_tasks: opt.experimental_max_number_of_batched_tasks,
             index_growth_amount: byte_unit::Byte::from_str("10GiB").unwrap().get_bytes() as usize,
             index_count: DEFAULT_INDEX_COUNT,
             instance_features,
@@ -277,15 +331,15 @@ fn import_dump(
     let mut dump_reader = dump::DumpReader::open(reader)?;
 
     if let Some(date) = dump_reader.date() {
-        log::info!(
-            "Importing a dump of meilisearch `{:?}` from the {}",
-            dump_reader.version(), // TODO: get the meilisearch version instead of the dump version
-            date
+        tracing::info!(
+            version = ?dump_reader.version(), // TODO: get the meilisearch version instead of the dump version
+            %date,
+            "Importing a dump of meilisearch"
         );
     } else {
-        log::info!(
-            "Importing a dump of meilisearch `{:?}`",
-            dump_reader.version(), // TODO: get the meilisearch version instead of the dump version
+        tracing::info!(
+            version = ?dump_reader.version(), // TODO: get the meilisearch version instead of the dump version
+            "Importing a dump of meilisearch",
         );
     }
 
@@ -319,7 +373,7 @@ fn import_dump(
     for index_reader in dump_reader.indexes()? {
         let mut index_reader = index_reader?;
         let metadata = index_reader.metadata();
-        log::info!("Importing index `{}`.", metadata.uid);
+        tracing::info!("Importing index `{}`.", metadata.uid);
 
         let date = Some((metadata.created_at, metadata.updated_at));
         let index = index_scheduler.create_raw_index(&metadata.uid, date)?;
@@ -333,14 +387,15 @@ fn import_dump(
         }
 
         // 4.2 Import the settings.
-        log::info!("Importing the settings.");
+        tracing::info!("Importing the settings.");
         let settings = index_reader.settings()?;
         apply_settings_to_builder(&settings, &mut builder);
-        builder.execute(|indexing_step| log::debug!("update: {:?}", indexing_step), || false)?;
+        builder
+            .execute(|indexing_step| tracing::debug!("update: {:?}", indexing_step), || false)?;
 
         // 4.3 Import the documents.
         // 4.3.1 We need to recreate the grenad+obkv format accepted by the index.
-        log::info!("Importing the documents.");
+        tracing::info!("Importing the documents.");
         let file = tempfile::tempfile()?;
         let mut builder = DocumentsBatchBuilder::new(BufWriter::new(file));
         for document in index_reader.documents()? {
@@ -362,15 +417,16 @@ fn import_dump(
                 update_method: IndexDocumentsMethod::ReplaceDocuments,
                 ..Default::default()
             },
-            |indexing_step| log::trace!("update: {:?}", indexing_step),
+            |indexing_step| tracing::trace!("update: {:?}", indexing_step),
             || false,
         )?;
 
         let (builder, user_result) = builder.add_documents(reader)?;
-        log::info!("{} documents found.", user_result?);
+        let user_result = user_result?;
+        tracing::info!(documents_found = user_result, "{} documents found.", user_result);
         builder.execute()?;
         wtxn.commit()?;
-        log::info!("All documents successfully imported.");
+        tracing::info!("All documents successfully imported.");
     }
 
     let mut index_scheduler_dump = index_scheduler.register_dumped_task()?;
@@ -388,6 +444,7 @@ pub fn configure_data(
     index_scheduler: Data<IndexScheduler>,
     auth: Data<AuthController>,
     opt: &Opt,
+    logs: LogRouteHandle,
     analytics: Arc<dyn Analytics>,
 ) {
     let http_payload_size_limit = opt.http_payload_size_limit.get_bytes() as usize;
@@ -395,6 +452,7 @@ pub fn configure_data(
         .app_data(index_scheduler)
         .app_data(auth)
         .app_data(web::Data::from(analytics))
+        .app_data(web::Data::new(logs))
         .app_data(
             web::JsonConfig::default()
                 .limit(http_payload_size_limit)

meilisearch/src/main.rs

@@ -1,6 +1,7 @@
 use std::env;
 use std::io::{stderr, Write};
 use std::path::PathBuf;
+use std::str::FromStr;
 use std::sync::Arc;
 
 use actix_web::http::KeepAlive;
@@ -9,33 +10,60 @@ use actix_web::HttpServer;
 use index_scheduler::IndexScheduler;
 use is_terminal::IsTerminal;
 use meilisearch::analytics::Analytics;
-use meilisearch::{analytics, create_app, prototype_name, setup_meilisearch, Opt};
+use meilisearch::{
+    analytics, create_app, prototype_name, setup_meilisearch, LogRouteHandle, LogRouteType, Opt,
+};
 use meilisearch_auth::{generate_master_key, AuthController, MASTER_KEY_MIN_SIZE};
+use mimalloc::MiMalloc;
 use termcolor::{Color, ColorChoice, ColorSpec, StandardStream, WriteColor};
+use tracing::level_filters::LevelFilter;
+use tracing_subscriber::layer::SubscriberExt as _;
+use tracing_subscriber::Layer;
 
 #[global_allocator]
-static ALLOC: mimalloc::MiMalloc = mimalloc::MiMalloc;
+static ALLOC: MiMalloc = MiMalloc;
+
+fn default_layer() -> LogRouteType {
+    None.with_filter(tracing_subscriber::filter::Targets::new().with_target("", LevelFilter::OFF))
+}
 
 /// does all the setup before meilisearch is launched
-fn setup(opt: &Opt) -> anyhow::Result<()> {
-    let mut log_builder = env_logger::Builder::new();
-    log_builder.parse_filters(&opt.log_level.to_string());
+fn setup(opt: &Opt) -> anyhow::Result<LogRouteHandle> {
+    let (route_layer, route_layer_handle) = tracing_subscriber::reload::Layer::new(default_layer());
+    let route_layer: tracing_subscriber::reload::Layer<_, _> = route_layer;
 
-    log_builder.init();
+    let subscriber = tracing_subscriber::registry().with(route_layer).with(
+        tracing_subscriber::fmt::layer()
+            .with_span_events(tracing_subscriber::fmt::format::FmtSpan::CLOSE)
+            .with_filter(
+                tracing_subscriber::filter::LevelFilter::from_str(&opt.log_level.to_string())
+                    .unwrap(),
+            ),
+    );
 
-    Ok(())
+    // set the subscriber as the default for the application
+    tracing::subscriber::set_global_default(subscriber).unwrap();
+
+    Ok(route_layer_handle)
+}
+
+fn on_panic(info: &std::panic::PanicInfo) {
+    let info = info.to_string().replace('\n', " ");
+    tracing::error!(%info);
 }
 
 #[actix_web::main]
 async fn main() -> anyhow::Result<()> {
     let (opt, config_read_from) = Opt::try_build()?;
 
+    std::panic::set_hook(Box::new(on_panic));
+
     anyhow::ensure!(
         !(cfg!(windows) && opt.experimental_reduce_indexing_memory_usage),
         "The `experimental-reduce-indexing-memory-usage` flag is not supported on Windows"
     );
 
-    setup(&opt)?;
+    let log_handle = setup(&opt)?;
 
     match (opt.env.as_ref(), &opt.master_key) {
         ("production", Some(master_key)) if master_key.len() < MASTER_KEY_MIN_SIZE => {
@@ -73,7 +101,7 @@ async fn main() -> anyhow::Result<()> {
 
     print_launch_resume(&opt, analytics.clone(), config_read_from);
 
-    run_http(index_scheduler, auth_controller, opt, analytics).await?;
+    run_http(index_scheduler, auth_controller, opt, log_handle, analytics).await?;
 
     Ok(())
 }
@@ -82,6 +110,7 @@ async fn run_http(
     index_scheduler: Arc<IndexScheduler>,
     auth_controller: Arc<AuthController>,
     opt: Opt,
+    logs: LogRouteHandle,
     analytics: Arc<dyn Analytics>,
 ) -> anyhow::Result<()> {
     let enable_dashboard = &opt.env == "development";
@@ -94,6 +123,7 @@ async fn run_http(
             index_scheduler.clone(),
             auth_controller.clone(),
             opt.clone(),
+            logs.clone(),
             analytics.clone(),
             enable_dashboard,
         )
@@ -103,7 +133,7 @@ async fn run_http(
     .keep_alive(KeepAlive::Os);
 
     if let Some(config) = opt_clone.get_ssl_config()? {
-        http_server.bind_rustls(opt_clone.http_addr, config)?.run().await?;
+        http_server.bind_rustls_021(opt_clone.http_addr, config)?.run().await?;
     } else {
         http_server.bind(&opt_clone.http_addr)?.run().await?;
     }

meilisearch/src/option.rs

@@ -20,7 +20,8 @@ use rustls::server::{
 use rustls::RootCertStore;
 use rustls_pemfile::{certs, pkcs8_private_keys, rsa_private_keys};
 use serde::{Deserialize, Serialize};
-use sysinfo::{RefreshKind, System, SystemExt};
+use sysinfo::{MemoryRefreshKind, RefreshKind, System};
+use url::Url;
 
 const POSSIBLE_ENV: [&str; 2] = ["development", "production"];
 
@@ -28,6 +29,8 @@ const MEILI_DB_PATH: &str = "MEILI_DB_PATH";
 const MEILI_HTTP_ADDR: &str = "MEILI_HTTP_ADDR";
 const MEILI_MASTER_KEY: &str = "MEILI_MASTER_KEY";
 const MEILI_ENV: &str = "MEILI_ENV";
+const MEILI_TASK_WEBHOOK_URL: &str = "MEILI_TASK_WEBHOOK_URL";
+const MEILI_TASK_WEBHOOK_AUTHORIZATION_HEADER: &str = "MEILI_TASK_WEBHOOK_AUTHORIZATION_HEADER";
 #[cfg(feature = "analytics")]
 const MEILI_NO_ANALYTICS: &str = "MEILI_NO_ANALYTICS";
 const MEILI_HTTP_PAYLOAD_SIZE_LIMIT: &str = "MEILI_HTTP_PAYLOAD_SIZE_LIMIT";
@@ -48,9 +51,12 @@ const MEILI_IGNORE_MISSING_DUMP: &str = "MEILI_IGNORE_MISSING_DUMP";
 const MEILI_IGNORE_DUMP_IF_DB_EXISTS: &str = "MEILI_IGNORE_DUMP_IF_DB_EXISTS";
 const MEILI_DUMP_DIR: &str = "MEILI_DUMP_DIR";
 const MEILI_LOG_LEVEL: &str = "MEILI_LOG_LEVEL";
+const MEILI_EXPERIMENTAL_ENABLE_LOGS_ROUTE: &str = "MEILI_EXPERIMENTAL_ENABLE_LOGS_ROUTE";
 const MEILI_EXPERIMENTAL_ENABLE_METRICS: &str = "MEILI_EXPERIMENTAL_ENABLE_METRICS";
 const MEILI_EXPERIMENTAL_REDUCE_INDEXING_MEMORY_USAGE: &str =
     "MEILI_EXPERIMENTAL_REDUCE_INDEXING_MEMORY_USAGE";
+const MEILI_EXPERIMENTAL_MAX_NUMBER_OF_BATCHED_TASKS: &str =
+    "MEILI_EXPERIMENTAL_MAX_NUMBER_OF_BATCHED_TASKS";
 
 const DEFAULT_CONFIG_FILE_PATH: &str = "./config.toml";
 const DEFAULT_DB_PATH: &str = "./data.ms";
@@ -154,6 +160,14 @@ pub struct Opt {
     #[serde(default = "default_env")]
     pub env: String,
 
+    /// Called whenever a task finishes so a third party can be notified.
+    #[clap(long, env = MEILI_TASK_WEBHOOK_URL)]
+    pub task_webhook_url: Option<Url>,
+
+    /// The Authorization header to send on the webhook URL whenever a task finishes so a third party can be notified.
+    #[clap(long, env = MEILI_TASK_WEBHOOK_AUTHORIZATION_HEADER)]
+    pub task_webhook_authorization_header: Option<String>,
+
     /// Deactivates Meilisearch's built-in telemetry when provided.
     ///
     /// Meilisearch automatically collects data from all instances that do not opt out using this flag.
@@ -296,11 +310,23 @@ pub struct Opt {
     #[serde(default)]
     pub experimental_enable_metrics: bool,
 
+    /// Experimental logs route feature. For more information, see: <https://github.com/orgs/meilisearch/discussions/721>
+    ///
+    /// Enables the log route on the `POST /logs/stream` endpoint and the `DELETE /logs/stream` to stop receiving logs.
+    #[clap(long, env = MEILI_EXPERIMENTAL_ENABLE_LOGS_ROUTE)]
+    #[serde(default)]
+    pub experimental_enable_logs_route: bool,
+
     /// Experimental RAM reduction during indexing, do not use in production, see: <https://github.com/meilisearch/product/discussions/652>
     #[clap(long, env = MEILI_EXPERIMENTAL_REDUCE_INDEXING_MEMORY_USAGE)]
     #[serde(default)]
     pub experimental_reduce_indexing_memory_usage: bool,
 
+    /// Experimentally reduces the maximum number of tasks that will be processed at once, see: <https://github.com/orgs/meilisearch/discussions/713>
+    #[clap(long, env = MEILI_EXPERIMENTAL_MAX_NUMBER_OF_BATCHED_TASKS, default_value_t = default_limit_batched_tasks())]
+    #[serde(default = "default_limit_batched_tasks")]
+    pub experimental_max_number_of_batched_tasks: usize,
+
     #[serde(flatten)]
     #[clap(flatten)]
     pub indexer_options: IndexerOpts,
@@ -368,9 +394,12 @@ impl Opt {
             http_addr,
             master_key,
             env,
+            task_webhook_url,
+            task_webhook_authorization_header,
             max_index_size: _,
             max_task_db_size: _,
             http_payload_size_limit,
+            experimental_max_number_of_batched_tasks,
             ssl_cert_path,
             ssl_key_path,
             ssl_auth_path,
@@ -392,8 +421,9 @@ impl Opt {
             config_file_path: _,
             #[cfg(feature = "analytics")]
             no_analytics,
-            experimental_enable_metrics: enable_metrics_route,
-            experimental_reduce_indexing_memory_usage: reduce_indexing_memory_usage,
+            experimental_enable_metrics,
+            experimental_enable_logs_route,
+            experimental_reduce_indexing_memory_usage,
         } = self;
         export_to_env_if_not_present(MEILI_DB_PATH, db_path);
         export_to_env_if_not_present(MEILI_HTTP_ADDR, http_addr);
@@ -401,6 +431,16 @@ impl Opt {
             export_to_env_if_not_present(MEILI_MASTER_KEY, master_key);
         }
         export_to_env_if_not_present(MEILI_ENV, env);
+        if let Some(task_webhook_url) = task_webhook_url {
+            export_to_env_if_not_present(MEILI_TASK_WEBHOOK_URL, task_webhook_url.to_string());
+        }
+        if let Some(task_webhook_authorization_header) = task_webhook_authorization_header {
+            export_to_env_if_not_present(
+                MEILI_TASK_WEBHOOK_AUTHORIZATION_HEADER,
+                task_webhook_authorization_header,
+            );
+        }
+
         #[cfg(feature = "analytics")]
         {
             export_to_env_if_not_present(MEILI_NO_ANALYTICS, no_analytics.to_string());
@@ -409,6 +449,10 @@ impl Opt {
             MEILI_HTTP_PAYLOAD_SIZE_LIMIT,
             http_payload_size_limit.to_string(),
         );
+        export_to_env_if_not_present(
+            MEILI_EXPERIMENTAL_MAX_NUMBER_OF_BATCHED_TASKS,
+            experimental_max_number_of_batched_tasks.to_string(),
+        );
         if let Some(ssl_cert_path) = ssl_cert_path {
             export_to_env_if_not_present(MEILI_SSL_CERT_PATH, ssl_cert_path);
         }
@@ -433,11 +477,15 @@ impl Opt {
         export_to_env_if_not_present(MEILI_LOG_LEVEL, log_level.to_string());
         export_to_env_if_not_present(
             MEILI_EXPERIMENTAL_ENABLE_METRICS,
-            enable_metrics_route.to_string(),
+            experimental_enable_metrics.to_string(),
+        );
+        export_to_env_if_not_present(
+            MEILI_EXPERIMENTAL_ENABLE_LOGS_ROUTE,
+            experimental_enable_logs_route.to_string(),
         );
         export_to_env_if_not_present(
             MEILI_EXPERIMENTAL_REDUCE_INDEXING_MEMORY_USAGE,
-            reduce_indexing_memory_usage.to_string(),
+            experimental_reduce_indexing_memory_usage.to_string(),
         );
         indexer_options.export_to_env();
     }
@@ -455,11 +503,11 @@ impl Opt {
                     }
                     if self.ssl_require_auth {
                         let verifier = AllowAnyAuthenticatedClient::new(client_auth_roots);
-                        config.with_client_cert_verifier(verifier)
+                        config.with_client_cert_verifier(Arc::from(verifier))
                     } else {
                         let verifier =
                             AllowAnyAnonymousOrAuthenticatedClient::new(client_auth_roots);
-                        config.with_client_cert_verifier(verifier)
+                        config.with_client_cert_verifier(Arc::from(verifier))
                     }
                 }
                 None => config.with_no_client_auth(),
@@ -489,7 +537,10 @@ impl Opt {
     }
 
     pub(crate) fn to_instance_features(&self) -> InstanceTogglableFeatures {
-        InstanceTogglableFeatures { metrics: self.experimental_enable_metrics }
+        InstanceTogglableFeatures {
+            metrics: self.experimental_enable_metrics,
+            logs_route: self.experimental_enable_logs_route,
+        }
     }
 }
 
@@ -598,8 +649,8 @@ impl MaxMemory {
 
 /// Returns the total amount of bytes available or `None` if this system isn't supported.
 fn total_memory_bytes() -> Option<u64> {
-    if System::IS_SUPPORTED {
-        let memory_kind = RefreshKind::new().with_memory();
+    if sysinfo::IS_SUPPORTED_SYSTEM {
+        let memory_kind = RefreshKind::new().with_memory(MemoryRefreshKind::new().with_ram());
         let mut system = System::new_with_specifics(memory_kind);
         system.refresh_memory();
         Some(system.total_memory())
@@ -727,6 +778,10 @@ fn default_http_payload_size_limit() -> Byte {
     Byte::from_str(DEFAULT_HTTP_PAYLOAD_SIZE_LIMIT).unwrap()
 }
 
+fn default_limit_batched_tasks() -> usize {
+    usize::MAX
+}
+
 fn default_snapshot_dir() -> PathBuf {
     PathBuf::from(DEFAULT_SNAPSHOT_DIR)
 }

meilisearch/src/routes/dump.rs

@@ -1,11 +1,11 @@
 use actix_web::web::Data;
 use actix_web::{web, HttpRequest, HttpResponse};
 use index_scheduler::IndexScheduler;
-use log::debug;
 use meilisearch_auth::AuthController;
 use meilisearch_types::error::ResponseError;
 use meilisearch_types::tasks::KindWithContent;
 use serde_json::json;
+use tracing::debug;
 
 use crate::analytics::Analytics;
 use crate::extractors::authentication::policies::*;
@@ -32,6 +32,6 @@ pub async fn create_dump(
     let task: SummarizedTaskView =
         tokio::task::spawn_blocking(move || index_scheduler.register(task)).await??.into();
 
-    debug!("returns: {:?}", task);
+    debug!(returns = ?task, "Create dump");
     Ok(HttpResponse::Accepted().json(task))
 }

meilisearch/src/routes/features.rs

@@ -3,11 +3,11 @@ use actix_web::{HttpRequest, HttpResponse};
 use deserr::actix_web::AwebJson;
 use deserr::Deserr;
 use index_scheduler::IndexScheduler;
-use log::debug;
 use meilisearch_types::deserr::DeserrJsonError;
 use meilisearch_types::error::ResponseError;
 use meilisearch_types::keys::actions;
 use serde_json::json;
+use tracing::debug;
 
 use crate::analytics::Analytics;
 use crate::extractors::authentication::policies::ActionPolicy;
@@ -33,23 +33,22 @@ async fn get_features(
     let features = index_scheduler.features();
 
     analytics.publish("Experimental features Seen".to_string(), json!(null), Some(&req));
-    debug!("returns: {:?}", features.runtime_features());
-    HttpResponse::Ok().json(features.runtime_features())
+    let features = features.runtime_features();
+    debug!(returns = ?features, "Get features");
+    HttpResponse::Ok().json(features)
 }
 
 #[derive(Debug, Deserr)]
 #[deserr(error = DeserrJsonError, rename_all = camelCase, deny_unknown_fields)]
 pub struct RuntimeTogglableFeatures {
-    #[deserr(default)]
-    pub score_details: Option<bool>,
     #[deserr(default)]
     pub vector_store: Option<bool>,
     #[deserr(default)]
     pub metrics: Option<bool>,
     #[deserr(default)]
-    pub export_puffin_reports: Option<bool>,
+    pub logs_route: Option<bool>,
     #[deserr(default)]
-    pub proximity_precision: Option<bool>,
+    pub export_puffin_reports: Option<bool>,
 }
 
 async fn patch_features(
@@ -62,44 +61,40 @@ async fn patch_features(
     analytics: Data<dyn Analytics>,
 ) -> Result<HttpResponse, ResponseError> {
     let features = index_scheduler.features();
+    debug!(parameters = ?new_features, "Patch features");
 
     let old_features = features.runtime_features();
     let new_features = meilisearch_types::features::RuntimeTogglableFeatures {
-        score_details: new_features.0.score_details.unwrap_or(old_features.score_details),
         vector_store: new_features.0.vector_store.unwrap_or(old_features.vector_store),
         metrics: new_features.0.metrics.unwrap_or(old_features.metrics),
+        logs_route: new_features.0.logs_route.unwrap_or(old_features.logs_route),
         export_puffin_reports: new_features
             .0
             .export_puffin_reports
             .unwrap_or(old_features.export_puffin_reports),
-        proximity_precision: new_features
-            .0
-            .proximity_precision
-            .unwrap_or(old_features.proximity_precision),
     };
 
     // explicitly destructure for analytics rather than using the `Serialize` implementation, because
     // the it renames to camelCase, which we don't want for analytics.
     // **Do not** ignore fields with `..` or `_` here, because we want to add them in the future.
     let meilisearch_types::features::RuntimeTogglableFeatures {
-        score_details,
         vector_store,
         metrics,
+        logs_route,
         export_puffin_reports,
-        proximity_precision,
     } = new_features;
 
     analytics.publish(
         "Experimental features Updated".to_string(),
         json!({
-            "score_details": score_details,
             "vector_store": vector_store,
             "metrics": metrics,
+            "logs_route": logs_route,
             "export_puffin_reports": export_puffin_reports,
-            "proximity_precision": proximity_precision,
         }),
         Some(&req),
     );
     index_scheduler.put_runtime_features(new_features)?;
+    debug!(returns = ?new_features, "Patch features");
     Ok(HttpResponse::Ok().json(new_features))
 }

meilisearch/src/routes/indexes/documents.rs

@@ -8,7 +8,6 @@ use deserr::actix_web::{AwebJson, AwebQueryParameter};
 use deserr::Deserr;
 use futures::StreamExt;
 use index_scheduler::IndexScheduler;
-use log::debug;
 use meilisearch_types::deserr::query_params::Param;
 use meilisearch_types::deserr::{DeserrJsonError, DeserrQueryParamError};
 use meilisearch_types::document_formats::{read_csv, read_json, read_ndjson, PayloadType};
@@ -28,6 +27,7 @@ use serde_json::Value;
 use tempfile::tempfile;
 use tokio::fs::File;
 use tokio::io::{AsyncSeekExt, AsyncWriteExt, BufWriter};
+use tracing::debug;
 
 use crate::analytics::{Analytics, DocumentDeletionKind, DocumentFetchKind};
 use crate::error::MeilisearchHttpError;
@@ -101,6 +101,7 @@ pub async fn get_document(
     analytics: web::Data<dyn Analytics>,
 ) -> Result<HttpResponse, ResponseError> {
     let DocumentParam { index_uid, document_id } = document_param.into_inner();
+    debug!(parameters = ?params, "Get document");
     let index_uid = IndexUid::try_from(index_uid)?;
 
     analytics.get_fetch_documents(&DocumentFetchKind::PerDocumentId, &req);
@@ -110,7 +111,7 @@ pub async fn get_document(
 
     let index = index_scheduler.index(&index_uid)?;
     let document = retrieve_document(&index, &document_id, attributes_to_retrieve)?;
-    debug!("returns: {:?}", document);
+    debug!(returns = ?document, "Get document");
     Ok(HttpResponse::Ok().json(document))
 }
 
@@ -131,7 +132,7 @@ pub async fn delete_document(
     };
     let task: SummarizedTaskView =
         tokio::task::spawn_blocking(move || index_scheduler.register(task)).await??.into();
-    debug!("returns: {:?}", task);
+    debug!(returns = ?task, "Delete document");
     Ok(HttpResponse::Accepted().json(task))
 }
 
@@ -168,9 +169,8 @@ pub async fn documents_by_query_post(
     req: HttpRequest,
     analytics: web::Data<dyn Analytics>,
 ) -> Result<HttpResponse, ResponseError> {
-    debug!("called with body: {:?}", body);
-
     let body = body.into_inner();
+    debug!(parameters = ?body, "Get documents POST");
 
     analytics.post_fetch_documents(
         &DocumentFetchKind::Normal {
@@ -191,7 +191,7 @@ pub async fn get_documents(
     req: HttpRequest,
     analytics: web::Data<dyn Analytics>,
 ) -> Result<HttpResponse, ResponseError> {
-    debug!("called with params: {:?}", params);
+    debug!(parameters = ?params, "Get documents GET");
 
     let BrowseQueryGet { limit, offset, fields, filter } = params.into_inner();
 
@@ -235,7 +235,7 @@ fn documents_by_query(
 
     let ret = PaginationView::new(offset, limit, total as usize, documents);
 
-    debug!("returns: {:?}", ret);
+    debug!(returns = ?ret, "Get documents");
     Ok(HttpResponse::Ok().json(ret))
 }
 
@@ -271,7 +271,7 @@ pub async fn replace_documents(
 ) -> Result<HttpResponse, ResponseError> {
     let index_uid = IndexUid::try_from(index_uid.into_inner())?;
 
-    debug!("called with params: {:?}", params);
+    debug!(parameters = ?params, "Replace documents");
     let params = params.into_inner();
 
     analytics.add_documents(&params, index_scheduler.index(&index_uid).is_err(), &req);
@@ -288,6 +288,7 @@ pub async fn replace_documents(
         allow_index_creation,
     )
     .await?;
+    debug!(returns = ?task, "Replace documents");
 
     Ok(HttpResponse::Accepted().json(task))
 }
@@ -302,8 +303,8 @@ pub async fn update_documents(
 ) -> Result<HttpResponse, ResponseError> {
     let index_uid = IndexUid::try_from(index_uid.into_inner())?;
 
-    debug!("called with params: {:?}", params);
     let params = params.into_inner();
+    debug!(parameters = ?params, "Update documents");
 
     analytics.update_documents(&params, index_scheduler.index(&index_uid).is_err(), &req);
 
@@ -319,6 +320,7 @@ pub async fn update_documents(
         allow_index_creation,
     )
     .await?;
+    debug!(returns = ?task, "Update documents");
 
     Ok(HttpResponse::Accepted().json(task))
 }
@@ -427,7 +429,10 @@ async fn document_addition(
                 Err(index_scheduler::Error::FileStore(file_store::Error::IoError(e)))
                     if e.kind() == ErrorKind::NotFound => {}
                 Err(e) => {
-                    log::warn!("Unknown error happened while deleting a malformed update file with uuid {uuid}: {e}");
+                    tracing::warn!(
+                        index_uuid = %uuid,
+                        "Unknown error happened while deleting a malformed update file: {e}"
+                    );
                 }
             }
             // We still want to return the original error to the end user.
@@ -453,7 +458,6 @@ async fn document_addition(
         }
     };
 
-    debug!("returns: {:?}", task);
     Ok(task.into())
 }
 
@@ -464,7 +468,7 @@ pub async fn delete_documents_batch(
     req: HttpRequest,
     analytics: web::Data<dyn Analytics>,
 ) -> Result<HttpResponse, ResponseError> {
-    debug!("called with params: {:?}", body);
+    debug!(parameters = ?body, "Delete documents by batch");
     let index_uid = IndexUid::try_from(index_uid.into_inner())?;
 
     analytics.delete_documents(DocumentDeletionKind::PerBatch, &req);
@@ -479,7 +483,7 @@ pub async fn delete_documents_batch(
     let task: SummarizedTaskView =
         tokio::task::spawn_blocking(move || index_scheduler.register(task)).await??.into();
 
-    debug!("returns: {:?}", task);
+    debug!(returns = ?task, "Delete documents by batch");
     Ok(HttpResponse::Accepted().json(task))
 }
 
@@ -497,7 +501,7 @@ pub async fn delete_documents_by_filter(
     req: HttpRequest,
     analytics: web::Data<dyn Analytics>,
 ) -> Result<HttpResponse, ResponseError> {
-    debug!("called with params: {:?}", body);
+    debug!(parameters = ?body, "Delete documents by filter");
     let index_uid = IndexUid::try_from(index_uid.into_inner())?;
     let index_uid = index_uid.into_inner();
     let filter = body.into_inner().filter;
@@ -515,7 +519,7 @@ pub async fn delete_documents_by_filter(
     let task: SummarizedTaskView =
         tokio::task::spawn_blocking(move || index_scheduler.register(task)).await??.into();
 
-    debug!("returns: {:?}", task);
+    debug!(returns = ?task, "Delete documents by filter");
     Ok(HttpResponse::Accepted().json(task))
 }
 
@@ -532,7 +536,7 @@ pub async fn clear_all_documents(
     let task: SummarizedTaskView =
         tokio::task::spawn_blocking(move || index_scheduler.register(task)).await??.into();
 
-    debug!("returns: {:?}", task);
+    debug!(returns = ?task, "Delete all documents");
     Ok(HttpResponse::Accepted().json(task))
 }
 

meilisearch/src/routes/indexes/facet_search.rs

@@ -2,20 +2,20 @@ use actix_web::web::Data;
 use actix_web::{web, HttpRequest, HttpResponse};
 use deserr::actix_web::AwebJson;
 use index_scheduler::IndexScheduler;
-use log::debug;
 use meilisearch_types::deserr::DeserrJsonError;
 use meilisearch_types::error::deserr_codes::*;
 use meilisearch_types::error::ResponseError;
 use meilisearch_types::index_uid::IndexUid;
 use serde_json::Value;
+use tracing::debug;
 
 use crate::analytics::{Analytics, FacetSearchAggregator};
 use crate::extractors::authentication::policies::*;
 use crate::extractors::authentication::GuardedData;
 use crate::search::{
-    add_search_rules, perform_facet_search, MatchingStrategy, SearchQuery, DEFAULT_CROP_LENGTH,
-    DEFAULT_CROP_MARKER, DEFAULT_HIGHLIGHT_POST_TAG, DEFAULT_HIGHLIGHT_PRE_TAG,
-    DEFAULT_SEARCH_LIMIT, DEFAULT_SEARCH_OFFSET,
+    add_search_rules, perform_facet_search, HybridQuery, MatchingStrategy, SearchQuery,
+    DEFAULT_CROP_LENGTH, DEFAULT_CROP_MARKER, DEFAULT_HIGHLIGHT_POST_TAG,
+    DEFAULT_HIGHLIGHT_PRE_TAG, DEFAULT_SEARCH_LIMIT, DEFAULT_SEARCH_OFFSET,
 };
 
 pub fn configure(cfg: &mut web::ServiceConfig) {
@@ -36,6 +36,8 @@ pub struct FacetSearchQuery {
     pub q: Option<String>,
     #[deserr(default, error = DeserrJsonError<InvalidSearchVector>)]
     pub vector: Option<Vec<f32>>,
+    #[deserr(default, error = DeserrJsonError<InvalidHybridQuery>)]
+    pub hybrid: Option<HybridQuery>,
     #[deserr(default, error = DeserrJsonError<InvalidSearchFilter>)]
     pub filter: Option<Value>,
     #[deserr(default, error = DeserrJsonError<InvalidSearchMatchingStrategy>, default)]
@@ -54,7 +56,7 @@ pub async fn search(
     let index_uid = IndexUid::try_from(index_uid.into_inner())?;
 
     let query = params.into_inner();
-    debug!("facet search called with params: {:?}", query);
+    debug!(parameters = ?query, "Facet search");
 
     let mut aggregate = FacetSearchAggregator::from_query(&query, &req);
 
@@ -81,7 +83,7 @@ pub async fn search(
 
     let search_result = search_result?;
 
-    debug!("returns: {:?}", search_result);
+    debug!(returns = ?search_result, "Facet search");
     Ok(HttpResponse::Ok().json(search_result))
 }
 
@@ -95,6 +97,7 @@ impl From<FacetSearchQuery> for SearchQuery {
             filter,
             matching_strategy,
             attributes_to_search_on,
+            hybrid,
         } = value;
 
         SearchQuery {
@@ -119,6 +122,7 @@ impl From<FacetSearchQuery> for SearchQuery {
             matching_strategy,
             vector,
             attributes_to_search_on,
+            hybrid,
         }
     }
 }

meilisearch/src/routes/indexes/mod.rs

@@ -5,7 +5,6 @@ use actix_web::{web, HttpRequest, HttpResponse};
 use deserr::actix_web::{AwebJson, AwebQueryParameter};
 use deserr::{DeserializeError, Deserr, ValuePointerRef};
 use index_scheduler::IndexScheduler;
-use log::debug;
 use meilisearch_types::deserr::query_params::Param;
 use meilisearch_types::deserr::{immutable_field_error, DeserrJsonError, DeserrQueryParamError};
 use meilisearch_types::error::deserr_codes::*;
@@ -16,6 +15,7 @@ use meilisearch_types::tasks::KindWithContent;
 use serde::Serialize;
 use serde_json::json;
 use time::OffsetDateTime;
+use tracing::debug;
 
 use super::{Pagination, SummarizedTaskView, PAGINATION_DEFAULT_LIMIT};
 use crate::analytics::Analytics;
@@ -93,6 +93,7 @@ pub async fn list_indexes(
     index_scheduler: GuardedData<ActionPolicy<{ actions::INDEXES_GET }>, Data<IndexScheduler>>,
     paginate: AwebQueryParameter<ListIndexes, DeserrQueryParamError>,
 ) -> Result<HttpResponse, ResponseError> {
+    debug!(parameters = ?paginate, "List indexes");
     let filters = index_scheduler.filters();
     let indexes: Vec<Option<IndexView>> =
         index_scheduler.try_for_each_index(|uid, index| -> Result<Option<IndexView>, _> {
@@ -105,7 +106,7 @@ pub async fn list_indexes(
     let indexes: Vec<IndexView> = indexes.into_iter().flatten().collect();
     let ret = paginate.as_pagination().auto_paginate_sized(indexes.into_iter());
 
-    debug!("returns: {:?}", ret);
+    debug!(returns = ?ret, "List indexes");
     Ok(HttpResponse::Ok().json(ret))
 }
 
@@ -124,6 +125,7 @@ pub async fn create_index(
     req: HttpRequest,
     analytics: web::Data<dyn Analytics>,
 ) -> Result<HttpResponse, ResponseError> {
+    debug!(parameters = ?body, "Create index");
     let IndexCreateRequest { primary_key, uid } = body.into_inner();
 
     let allow_index_creation = index_scheduler.filters().allow_index_creation(&uid);
@@ -137,6 +139,7 @@ pub async fn create_index(
         let task = KindWithContent::IndexCreation { index_uid: uid.to_string(), primary_key };
         let task: SummarizedTaskView =
             tokio::task::spawn_blocking(move || index_scheduler.register(task)).await??.into();
+        debug!(returns = ?task, "Create index");
 
         Ok(HttpResponse::Accepted().json(task))
     } else {
@@ -177,7 +180,7 @@ pub async fn get_index(
     let index = index_scheduler.index(&index_uid)?;
     let index_view = IndexView::new(index_uid.into_inner(), &index)?;
 
-    debug!("returns: {:?}", index_view);
+    debug!(returns = ?index_view, "Get index");
 
     Ok(HttpResponse::Ok().json(index_view))
 }
@@ -189,7 +192,7 @@ pub async fn update_index(
     req: HttpRequest,
     analytics: web::Data<dyn Analytics>,
 ) -> Result<HttpResponse, ResponseError> {
-    debug!("called with params: {:?}", body);
+    debug!(parameters = ?body, "Update index");
     let index_uid = IndexUid::try_from(index_uid.into_inner())?;
     let body = body.into_inner();
     analytics.publish(
@@ -206,7 +209,7 @@ pub async fn update_index(
     let task: SummarizedTaskView =
         tokio::task::spawn_blocking(move || index_scheduler.register(task)).await??.into();
 
-    debug!("returns: {:?}", task);
+    debug!(returns = ?task, "Update index");
     Ok(HttpResponse::Accepted().json(task))
 }
 
@@ -218,6 +221,7 @@ pub async fn delete_index(
     let task = KindWithContent::IndexDeletion { index_uid: index_uid.into_inner() };
     let task: SummarizedTaskView =
         tokio::task::spawn_blocking(move || index_scheduler.register(task)).await??.into();
+    debug!(returns = ?task, "Delete index");
 
     Ok(HttpResponse::Accepted().json(task))
 }
@@ -255,6 +259,6 @@ pub async fn get_index_stats(
 
     let stats = IndexStats::from(index_scheduler.index_stats(&index_uid)?);
 
-    debug!("returns: {:?}", stats);
+    debug!(returns = ?stats, "Get index stats");
     Ok(HttpResponse::Ok().json(stats))
 }

meilisearch/src/routes/indexes/search.rs

@@ -2,23 +2,25 @@ use actix_web::web::Data;
 use actix_web::{web, HttpRequest, HttpResponse};
 use deserr::actix_web::{AwebJson, AwebQueryParameter};
 use index_scheduler::IndexScheduler;
-use log::debug;
 use meilisearch_types::deserr::query_params::Param;
 use meilisearch_types::deserr::{DeserrJsonError, DeserrQueryParamError};
 use meilisearch_types::error::deserr_codes::*;
 use meilisearch_types::error::ResponseError;
 use meilisearch_types::index_uid::IndexUid;
+use meilisearch_types::milli;
+use meilisearch_types::milli::vector::DistributionShift;
 use meilisearch_types::serde_cs::vec::CS;
 use serde_json::Value;
+use tracing::{debug, warn};
 
 use crate::analytics::{Analytics, SearchAggregator};
 use crate::extractors::authentication::policies::*;
 use crate::extractors::authentication::GuardedData;
 use crate::extractors::sequential_extractor::SeqHandler;
 use crate::search::{
-    add_search_rules, perform_search, MatchingStrategy, SearchQuery, DEFAULT_CROP_LENGTH,
-    DEFAULT_CROP_MARKER, DEFAULT_HIGHLIGHT_POST_TAG, DEFAULT_HIGHLIGHT_PRE_TAG,
-    DEFAULT_SEARCH_LIMIT, DEFAULT_SEARCH_OFFSET,
+    add_search_rules, perform_search, HybridQuery, MatchingStrategy, SearchQuery, SemanticRatio,
+    DEFAULT_CROP_LENGTH, DEFAULT_CROP_MARKER, DEFAULT_HIGHLIGHT_POST_TAG,
+    DEFAULT_HIGHLIGHT_PRE_TAG, DEFAULT_SEARCH_LIMIT, DEFAULT_SEARCH_OFFSET, DEFAULT_SEMANTIC_RATIO,
 };
 
 pub fn configure(cfg: &mut web::ServiceConfig) {
@@ -74,6 +76,31 @@ pub struct SearchQueryGet {
     matching_strategy: MatchingStrategy,
     #[deserr(default, error = DeserrQueryParamError<InvalidSearchAttributesToSearchOn>)]
     pub attributes_to_search_on: Option<CS<String>>,
+    #[deserr(default, error = DeserrQueryParamError<InvalidEmbedder>)]
+    pub hybrid_embedder: Option<String>,
+    #[deserr(default, error = DeserrQueryParamError<InvalidSearchSemanticRatio>)]
+    pub hybrid_semantic_ratio: Option<SemanticRatioGet>,
+}
+
+#[derive(Debug, Clone, Copy, Default, PartialEq, deserr::Deserr)]
+#[deserr(try_from(String) = TryFrom::try_from -> InvalidSearchSemanticRatio)]
+pub struct SemanticRatioGet(SemanticRatio);
+
+impl std::convert::TryFrom<String> for SemanticRatioGet {
+    type Error = InvalidSearchSemanticRatio;
+
+    fn try_from(s: String) -> Result<Self, Self::Error> {
+        let f: f32 = s.parse().map_err(|_| InvalidSearchSemanticRatio)?;
+        Ok(SemanticRatioGet(SemanticRatio::try_from(f)?))
+    }
+}
+
+impl std::ops::Deref for SemanticRatioGet {
+    type Target = SemanticRatio;
+
+    fn deref(&self) -> &Self::Target {
+        &self.0
+    }
 }
 
 impl From<SearchQueryGet> for SearchQuery {
@@ -86,6 +113,20 @@ impl From<SearchQueryGet> for SearchQuery {
             None => None,
         };
 
+        let hybrid = match (other.hybrid_embedder, other.hybrid_semantic_ratio) {
+            (None, None) => None,
+            (None, Some(semantic_ratio)) => {
+                Some(HybridQuery { semantic_ratio: *semantic_ratio, embedder: None })
+            }
+            (Some(embedder), None) => Some(HybridQuery {
+                semantic_ratio: DEFAULT_SEMANTIC_RATIO(),
+                embedder: Some(embedder),
+            }),
+            (Some(embedder), Some(semantic_ratio)) => {
+                Some(HybridQuery { semantic_ratio: *semantic_ratio, embedder: Some(embedder) })
+            }
+        };
+
         Self {
             q: other.q,
             vector: other.vector.map(CS::into_inner),
@@ -108,6 +149,7 @@ impl From<SearchQueryGet> for SearchQuery {
             crop_marker: other.crop_marker,
             matching_strategy: other.matching_strategy,
             attributes_to_search_on: other.attributes_to_search_on.map(|o| o.into_iter().collect()),
+            hybrid,
         }
     }
 }
@@ -144,7 +186,7 @@ pub async fn search_with_url_query(
     req: HttpRequest,
     analytics: web::Data<dyn Analytics>,
 ) -> Result<HttpResponse, ResponseError> {
-    debug!("called with params: {:?}", params);
+    debug!(parameters = ?params, "Search get");
     let index_uid = IndexUid::try_from(index_uid.into_inner())?;
 
     let mut query: SearchQuery = params.into_inner().into();
@@ -158,8 +200,12 @@ pub async fn search_with_url_query(
 
     let index = index_scheduler.index(&index_uid)?;
     let features = index_scheduler.features();
+
+    let distribution = embed(&mut query, index_scheduler.get_ref(), &index).await?;
+
     let search_result =
-        tokio::task::spawn_blocking(move || perform_search(&index, query, features)).await?;
+        tokio::task::spawn_blocking(move || perform_search(&index, query, features, distribution))
+            .await?;
     if let Ok(ref search_result) = search_result {
         aggregate.succeed(search_result);
     }
@@ -167,7 +213,7 @@ pub async fn search_with_url_query(
 
     let search_result = search_result?;
 
-    debug!("returns: {:?}", search_result);
+    debug!(returns = ?search_result, "Search get");
     Ok(HttpResponse::Ok().json(search_result))
 }
 
@@ -181,7 +227,7 @@ pub async fn search_with_post(
     let index_uid = IndexUid::try_from(index_uid.into_inner())?;
 
     let mut query = params.into_inner();
-    debug!("search called with params: {:?}", query);
+    debug!(parameters = ?query, "Search post");
 
     // Tenant token search_rules.
     if let Some(search_rules) = index_scheduler.filters().get_index_search_rules(&index_uid) {
@@ -193,8 +239,12 @@ pub async fn search_with_post(
     let index = index_scheduler.index(&index_uid)?;
 
     let features = index_scheduler.features();
+
+    let distribution = embed(&mut query, index_scheduler.get_ref(), &index).await?;
+
     let search_result =
-        tokio::task::spawn_blocking(move || perform_search(&index, query, features)).await?;
+        tokio::task::spawn_blocking(move || perform_search(&index, query, features, distribution))
+            .await?;
     if let Ok(ref search_result) = search_result {
         aggregate.succeed(search_result);
     }
@@ -202,10 +252,84 @@ pub async fn search_with_post(
 
     let search_result = search_result?;
 
-    debug!("returns: {:?}", search_result);
+    debug!(returns = ?search_result, "Search post");
     Ok(HttpResponse::Ok().json(search_result))
 }
 
+pub async fn embed(
+    query: &mut SearchQuery,
+    index_scheduler: &IndexScheduler,
+    index: &milli::Index,
+) -> Result<Option<DistributionShift>, ResponseError> {
+    match (&query.hybrid, &query.vector, &query.q) {
+        (Some(HybridQuery { semantic_ratio: _, embedder }), None, Some(q))
+            if !q.trim().is_empty() =>
+        {
+            let embedder_configs = index.embedding_configs(&index.read_txn()?)?;
+            let embedders = index_scheduler.embedders(embedder_configs)?;
+
+            let embedder = if let Some(embedder_name) = embedder {
+                embedders.get(embedder_name)
+            } else {
+                embedders.get_default()
+            };
+
+            let embedder = embedder
+                .ok_or(milli::UserError::InvalidEmbedder("default".to_owned()))
+                .map_err(milli::Error::from)?
+                .0;
+
+            let distribution = embedder.distribution();
+
+            let embeddings = embedder
+                .embed(vec![q.to_owned()])
+                .await
+                .map_err(milli::vector::Error::from)
+                .map_err(milli::Error::from)?
+                .pop()
+                .expect("No vector returned from embedding");
+
+            if embeddings.iter().nth(1).is_some() {
+                warn!("Ignoring embeddings past the first one in long search query");
+                query.vector = Some(embeddings.iter().next().unwrap().to_vec());
+            } else {
+                query.vector = Some(embeddings.into_inner());
+            }
+            Ok(distribution)
+        }
+        (Some(hybrid), vector, _) => {
+            let embedder_configs = index.embedding_configs(&index.read_txn()?)?;
+            let embedders = index_scheduler.embedders(embedder_configs)?;
+
+            let embedder = if let Some(embedder_name) = &hybrid.embedder {
+                embedders.get(embedder_name)
+            } else {
+                embedders.get_default()
+            };
+
+            let embedder = embedder
+                .ok_or(milli::UserError::InvalidEmbedder("default".to_owned()))
+                .map_err(milli::Error::from)?
+                .0;
+
+            if let Some(vector) = vector {
+                if vector.len() != embedder.dimensions() {
+                    return Err(meilisearch_types::milli::Error::UserError(
+                        meilisearch_types::milli::UserError::InvalidVectorDimensions {
+                            expected: embedder.dimensions(),
+                            found: vector.len(),
+                        },
+                    )
+                    .into());
+                }
+            }
+
+            Ok(embedder.distribution())
+        }
+        _ => Ok(None),
+    }
+}
+
 #[cfg(test)]
 mod test {
     use super::*;

meilisearch/src/routes/indexes/settings.rs

@@ -2,14 +2,15 @@ use actix_web::web::Data;
 use actix_web::{web, HttpRequest, HttpResponse};
 use deserr::actix_web::AwebJson;
 use index_scheduler::IndexScheduler;
-use log::debug;
 use meilisearch_types::deserr::DeserrJsonError;
 use meilisearch_types::error::ResponseError;
 use meilisearch_types::facet_values_sort::FacetValuesSort;
 use meilisearch_types::index_uid::IndexUid;
+use meilisearch_types::milli::update::Setting;
 use meilisearch_types::settings::{settings, RankingRuleView, Settings, Unchecked};
 use meilisearch_types::tasks::KindWithContent;
 use serde_json::json;
+use tracing::debug;
 
 use crate::analytics::Analytics;
 use crate::extractors::authentication::policies::*;
@@ -23,12 +24,12 @@ macro_rules! make_setting_route {
             use actix_web::web::Data;
             use actix_web::{web, HttpRequest, HttpResponse, Resource};
             use index_scheduler::IndexScheduler;
-            use log::debug;
             use meilisearch_types::error::ResponseError;
             use meilisearch_types::index_uid::IndexUid;
             use meilisearch_types::milli::update::Setting;
             use meilisearch_types::settings::{settings, Settings};
             use meilisearch_types::tasks::KindWithContent;
+            use tracing::debug;
             use $crate::analytics::Analytics;
             use $crate::extractors::authentication::policies::*;
             use $crate::extractors::authentication::GuardedData;
@@ -60,7 +61,7 @@ macro_rules! make_setting_route {
                         .await??
                         .into();
 
-                debug!("returns: {:?}", task);
+                debug!(returns = ?task, "Delete settings");
                 Ok(HttpResponse::Accepted().json(task))
             }
 
@@ -77,6 +78,7 @@ macro_rules! make_setting_route {
                 let index_uid = IndexUid::try_from(index_uid.into_inner())?;
 
                 let body = body.into_inner();
+                debug!(parameters = ?body, "Update settings");
 
                 #[allow(clippy::redundant_closure_call)]
                 $analytics(&body, &req);
@@ -89,6 +91,11 @@ macro_rules! make_setting_route {
                     ..Default::default()
                 };
 
+                let new_settings = $crate::routes::indexes::settings::validate_settings(
+                    new_settings,
+                    &index_scheduler,
+                )?;
+
                 let allow_index_creation =
                     index_scheduler.filters().allow_index_creation(&index_uid);
 
@@ -103,7 +110,7 @@ macro_rules! make_setting_route {
                         .await??
                         .into();
 
-                debug!("returns: {:?}", task);
+                debug!(returns = ?task, "Update settings");
                 Ok(HttpResponse::Accepted().json(task))
             }
 
@@ -120,7 +127,7 @@ macro_rules! make_setting_route {
                 let rtxn = index.read_txn()?;
                 let settings = settings(&index, &rtxn)?;
 
-                debug!("returns: {:?}", settings);
+                debug!(returns = ?settings, "Update settings");
                 let mut json = serde_json::json!(&settings);
                 let val = json[$camelcase_attr].take();
 
@@ -452,6 +459,7 @@ make_setting_route!(
             json!({
                 "proximity_precision": {
                     "set": precision.is_some(),
+                    "value": precision.unwrap_or_default(),
                 }
             }),
             Some(req),
@@ -545,6 +553,67 @@ make_setting_route!(
     }
 );
 
+make_setting_route!(
+    "/embedders",
+    patch,
+    std::collections::BTreeMap<String, Setting<meilisearch_types::milli::vector::settings::EmbeddingSettings>>,
+    meilisearch_types::deserr::DeserrJsonError<
+        meilisearch_types::error::deserr_codes::InvalidSettingsEmbedders,
+    >,
+    embedders,
+    "embedders",
+    analytics,
+    |setting: &Option<std::collections::BTreeMap<String, Setting<meilisearch_types::milli::vector::settings::EmbeddingSettings>>>, req: &HttpRequest| {
+
+
+        analytics.publish(
+            "Embedders Updated".to_string(),
+            serde_json::json!({"embedders": crate::routes::indexes::settings::embedder_analytics(setting.as_ref())}),
+            Some(req),
+        );
+    }
+);
+
+fn embedder_analytics(
+    setting: Option<
+        &std::collections::BTreeMap<
+            String,
+            Setting<meilisearch_types::milli::vector::settings::EmbeddingSettings>,
+        >,
+    >,
+) -> serde_json::Value {
+    let mut sources = std::collections::HashSet::new();
+
+    if let Some(s) = &setting {
+        for source in s
+            .values()
+            .filter_map(|config| config.clone().set())
+            .filter_map(|config| config.source.set())
+        {
+            use meilisearch_types::milli::vector::settings::EmbedderSource;
+            match source {
+                EmbedderSource::OpenAi => sources.insert("openAi"),
+                EmbedderSource::HuggingFace => sources.insert("huggingFace"),
+                EmbedderSource::UserProvided => sources.insert("userProvided"),
+            };
+        }
+    };
+
+    let document_template_used = setting.as_ref().map(|map| {
+        map.values()
+            .filter_map(|config| config.clone().set())
+            .any(|config| config.document_template.set().is_some())
+    });
+
+    json!(
+        {
+            "total": setting.as_ref().map(|s| s.len()),
+            "sources": sources,
+            "document_template_used": document_template_used,
+        }
+    )
+}
+
 macro_rules! generate_configure {
     ($($mod:ident),*) => {
         pub fn configure(cfg: &mut web::ServiceConfig) {
@@ -574,7 +643,8 @@ generate_configure!(
     ranking_rules,
     typo_tolerance,
     pagination,
-    faceting
+    faceting,
+    embedders
 );
 
 pub async fn update_all(
@@ -587,6 +657,8 @@ pub async fn update_all(
     let index_uid = IndexUid::try_from(index_uid.into_inner())?;
 
     let new_settings = body.into_inner();
+    debug!(parameters = ?new_settings, "Update all settings");
+    let new_settings = validate_settings(new_settings, &index_scheduler)?;
 
     analytics.publish(
         "Settings Updated".to_string(),
@@ -620,7 +692,8 @@ pub async fn update_all(
                 "set": new_settings.distinct_attribute.as_ref().set().is_some()
             },
             "proximity_precision": {
-                "set": new_settings.proximity_precision.as_ref().set().is_some()
+                "set": new_settings.proximity_precision.as_ref().set().is_some(),
+                "value": new_settings.proximity_precision.as_ref().set().copied().unwrap_or_default()
             },
             "typo_tolerance": {
                 "enabled": new_settings.typo_tolerance
@@ -681,6 +754,7 @@ pub async fn update_all(
             "synonyms": {
                 "total": new_settings.synonyms.as_ref().set().map(|synonyms| synonyms.len()),
             },
+            "embedders": crate::routes::indexes::settings::embedder_analytics(new_settings.embedders.as_ref().set())
         }),
         Some(&req),
     );
@@ -696,7 +770,7 @@ pub async fn update_all(
     let task: SummarizedTaskView =
         tokio::task::spawn_blocking(move || index_scheduler.register(task)).await??.into();
 
-    debug!("returns: {:?}", task);
+    debug!(returns = ?task, "Update all settings");
     Ok(HttpResponse::Accepted().json(task))
 }
 
@@ -709,7 +783,7 @@ pub async fn get_all(
     let index = index_scheduler.index(&index_uid)?;
     let rtxn = index.read_txn()?;
     let new_settings = settings(&index, &rtxn)?;
-    debug!("returns: {:?}", new_settings);
+    debug!(returns = ?new_settings, "Get all settings");
     Ok(HttpResponse::Ok().json(new_settings))
 }
 
@@ -732,6 +806,16 @@ pub async fn delete_all(
     let task: SummarizedTaskView =
         tokio::task::spawn_blocking(move || index_scheduler.register(task)).await??.into();
 
-    debug!("returns: {:?}", task);
+    debug!(returns = ?task, "Delete all settings");
     Ok(HttpResponse::Accepted().json(task))
 }
+
+fn validate_settings(
+    settings: Settings<Unchecked>,
+    index_scheduler: &IndexScheduler,
+) -> Result<Settings<Unchecked>, ResponseError> {
+    if matches!(settings.embedders, Setting::Set(_)) {
+        index_scheduler.features().check_vector("Passing `embedders` in settings")?
+    }
+    Ok(settings.validate()?)
+}

meilisearch/src/routes/logs.rs

@@ -0,0 +1,281 @@
+use std::convert::Infallible;
+use std::io::Write;
+use std::ops::ControlFlow;
+use std::pin::Pin;
+use std::str::FromStr;
+use std::sync::Arc;
+
+use actix_web::web::{Bytes, Data};
+use actix_web::{web, HttpResponse};
+use deserr::actix_web::AwebJson;
+use deserr::{DeserializeError, Deserr, ErrorKind, MergeWithError, ValuePointerRef};
+use futures_util::Stream;
+use index_scheduler::IndexScheduler;
+use meilisearch_types::deserr::DeserrJsonError;
+use meilisearch_types::error::deserr_codes::*;
+use meilisearch_types::error::{Code, ResponseError};
+use tokio::sync::mpsc;
+use tracing_subscriber::filter::Targets;
+use tracing_subscriber::Layer;
+
+use crate::error::MeilisearchHttpError;
+use crate::extractors::authentication::policies::*;
+use crate::extractors::authentication::GuardedData;
+use crate::extractors::sequential_extractor::SeqHandler;
+use crate::LogRouteHandle;
+
+pub fn configure(cfg: &mut web::ServiceConfig) {
+    cfg.service(
+        web::resource("stream")
+            .route(web::post().to(SeqHandler(get_logs)))
+            .route(web::delete().to(SeqHandler(cancel_logs))),
+    );
+}
+
+#[derive(Debug, Default, Clone, Copy, Deserr, PartialEq, Eq)]
+#[deserr(rename_all = camelCase)]
+pub enum LogMode {
+    #[default]
+    Human,
+    Profile,
+}
+
+/// Simple wrapper around the `Targets` from `tracing_subscriber` to implement `MergeWithError` on it.
+#[derive(Clone, Debug)]
+struct MyTargets(Targets);
+
+/// Simple wrapper around the `ParseError` from `tracing_subscriber` to implement `MergeWithError` on it.
+#[derive(Debug, thiserror::Error)]
+enum MyParseError {
+    #[error(transparent)]
+    ParseError(#[from] tracing_subscriber::filter::ParseError),
+    #[error(
+        "Empty string is not a valid target. If you want to get no logs use `OFF`. Usage: `info`, `meilisearch=info`, or you can write multiple filters in one target: `index_scheduler=info,milli=trace`"
+    )]
+    Example,
+}
+
+impl FromStr for MyTargets {
+    type Err = MyParseError;
+
+    fn from_str(s: &str) -> Result<Self, Self::Err> {
+        if s.is_empty() {
+            Err(MyParseError::Example)
+        } else {
+            Ok(MyTargets(Targets::from_str(s).map_err(MyParseError::ParseError)?))
+        }
+    }
+}
+
+impl MergeWithError<MyParseError> for DeserrJsonError<BadRequest> {
+    fn merge(
+        _self_: Option<Self>,
+        other: MyParseError,
+        merge_location: ValuePointerRef,
+    ) -> ControlFlow<Self, Self> {
+        Self::error::<Infallible>(
+            None,
+            ErrorKind::Unexpected { msg: other.to_string() },
+            merge_location,
+        )
+    }
+}
+
+#[derive(Debug, Deserr)]
+#[deserr(error = DeserrJsonError, rename_all = camelCase, deny_unknown_fields, validate = validate_get_logs -> DeserrJsonError<InvalidSettingsTypoTolerance>)]
+pub struct GetLogs {
+    #[deserr(default = "info".parse().unwrap(), try_from(&String) = MyTargets::from_str -> DeserrJsonError<BadRequest>)]
+    target: MyTargets,
+
+    #[deserr(default, error = DeserrJsonError<BadRequest>)]
+    mode: LogMode,
+
+    #[deserr(default = false, error = DeserrJsonError<BadRequest>)]
+    profile_memory: bool,
+}
+
+fn validate_get_logs<E: DeserializeError>(
+    logs: GetLogs,
+    location: ValuePointerRef,
+) -> Result<GetLogs, E> {
+    if logs.profile_memory && logs.mode != LogMode::Profile {
+        Err(deserr::take_cf_content(E::error::<Infallible>(
+            None,
+            ErrorKind::Unexpected {
+                msg: format!("`profile_memory` can only be used while profiling code and is not compatible with the {:?} mode.", logs.mode),
+            },
+            location,
+        )))
+    } else {
+        Ok(logs)
+    }
+}
+
+struct LogWriter {
+    sender: mpsc::UnboundedSender<Vec<u8>>,
+}
+
+impl Write for LogWriter {
+    fn write(&mut self, buf: &[u8]) -> std::io::Result<usize> {
+        self.sender.send(buf.to_vec()).map_err(std::io::Error::other)?;
+        Ok(buf.len())
+    }
+
+    fn flush(&mut self) -> std::io::Result<()> {
+        Ok(())
+    }
+}
+
+struct HandleGuard {
+    /// We need to keep an handle on the logs to make it available again when the streamer is dropped
+    logs: Arc<LogRouteHandle>,
+}
+
+impl Drop for HandleGuard {
+    fn drop(&mut self) {
+        if let Err(e) = self.logs.modify(|layer| *layer.inner_mut() = None) {
+            tracing::error!("Could not free the logs route: {e}");
+        }
+    }
+}
+
+fn byte_stream(
+    receiver: mpsc::UnboundedReceiver<Vec<u8>>,
+    guard: HandleGuard,
+) -> impl futures_util::Stream<Item = Result<Bytes, ResponseError>> {
+    futures_util::stream::unfold((receiver, guard), move |(mut receiver, guard)| async move {
+        let vec = receiver.recv().await;
+
+        vec.map(From::from).map(Ok).map(|a| (a, (receiver, guard)))
+    })
+}
+
+type PinnedByteStream = Pin<Box<dyn Stream<Item = Result<Bytes, ResponseError>>>>;
+
+fn make_layer<
+    S: tracing::Subscriber + for<'span> tracing_subscriber::registry::LookupSpan<'span>,
+>(
+    opt: &GetLogs,
+    logs: Data<LogRouteHandle>,
+) -> (Box<dyn Layer<S> + Send + Sync>, PinnedByteStream) {
+    let guard = HandleGuard { logs: logs.into_inner() };
+    match opt.mode {
+        LogMode::Human => {
+            let (sender, receiver) = tokio::sync::mpsc::unbounded_channel();
+
+            let fmt_layer = tracing_subscriber::fmt::layer()
+                .with_writer(move || LogWriter { sender: sender.clone() })
+                .with_span_events(tracing_subscriber::fmt::format::FmtSpan::ACTIVE);
+
+            let stream = byte_stream(receiver, guard);
+            (Box::new(fmt_layer) as Box<dyn Layer<S> + Send + Sync>, Box::pin(stream))
+        }
+        LogMode::Profile => {
+            let (trace, layer) = tracing_trace::Trace::new(opt.profile_memory);
+
+            let stream = entry_stream(trace, guard);
+
+            (Box::new(layer) as Box<dyn Layer<S> + Send + Sync>, Box::pin(stream))
+        }
+    }
+}
+
+fn entry_stream(
+    trace: tracing_trace::Trace,
+    guard: HandleGuard,
+) -> impl Stream<Item = Result<Bytes, ResponseError>> {
+    let receiver = trace.into_receiver();
+    let entry_buf = Vec::new();
+
+    futures_util::stream::unfold(
+        (receiver, entry_buf, guard),
+        move |(mut receiver, mut entry_buf, guard)| async move {
+            let mut bytes = Vec::new();
+
+            while bytes.len() < 8192 {
+                entry_buf.clear();
+
+                let Ok(count) = tokio::time::timeout(
+                    std::time::Duration::from_secs(1),
+                    receiver.recv_many(&mut entry_buf, 100),
+                )
+                .await
+                else {
+                    break;
+                };
+
+                if count == 0 {
+                    if !bytes.is_empty() {
+                        break;
+                    }
+
+                    // channel closed, exit
+                    return None;
+                }
+
+                for entry in &entry_buf {
+                    if let Err(error) = serde_json::to_writer(&mut bytes, entry) {
+                        tracing::error!(
+                            error = &error as &dyn std::error::Error,
+                            "deserializing entry"
+                        );
+                        return Some((
+                            Err(ResponseError::from_msg(
+                                format!("error deserializing entry: {error}"),
+                                Code::Internal,
+                            )),
+                            (receiver, entry_buf, guard),
+                        ));
+                    }
+                }
+            }
+
+            Some((Ok(bytes.into()), (receiver, entry_buf, guard)))
+        },
+    )
+}
+
+pub async fn get_logs(
+    index_scheduler: GuardedData<ActionPolicy<{ actions::METRICS_GET }>, Data<IndexScheduler>>,
+    logs: Data<LogRouteHandle>,
+    body: AwebJson<GetLogs, DeserrJsonError>,
+) -> Result<HttpResponse, ResponseError> {
+    index_scheduler.features().check_logs_route()?;
+
+    let opt = body.into_inner();
+    let mut stream = None;
+
+    logs.modify(|layer| match layer.inner_mut() {
+        None => {
+            // there is no one getting logs
+            *layer.filter_mut() = opt.target.0.clone();
+            let (new_layer, new_stream) = make_layer(&opt, logs.clone());
+
+            *layer.inner_mut() = Some(new_layer);
+            stream = Some(new_stream);
+        }
+        Some(_) => {
+            // there is already someone getting logs
+        }
+    })
+    .unwrap();
+
+    if let Some(stream) = stream {
+        Ok(HttpResponse::Ok().streaming(stream))
+    } else {
+        Err(MeilisearchHttpError::AlreadyUsedLogRoute.into())
+    }
+}
+
+pub async fn cancel_logs(
+    index_scheduler: GuardedData<ActionPolicy<{ actions::METRICS_GET }>, Data<IndexScheduler>>,
+    logs: Data<LogRouteHandle>,
+) -> Result<HttpResponse, ResponseError> {
+    index_scheduler.features().check_logs_route()?;
+
+    if let Err(e) = logs.modify(|layer| *layer.inner_mut() = None) {
+        tracing::error!("Could not free the logs route: {e}");
+    }
+
+    Ok(HttpResponse::NoContent().finish())
+}

meilisearch/src/routes/mod.rs

@@ -3,7 +3,6 @@ use std::collections::BTreeMap;
 use actix_web::web::Data;
 use actix_web::{web, HttpRequest, HttpResponse};
 use index_scheduler::IndexScheduler;
-use log::debug;
 use meilisearch_auth::AuthController;
 use meilisearch_types::error::ResponseError;
 use meilisearch_types::settings::{Settings, Unchecked};
@@ -11,6 +10,7 @@ use meilisearch_types::tasks::{Kind, Status, Task, TaskId};
 use serde::{Deserialize, Serialize};
 use serde_json::json;
 use time::OffsetDateTime;
+use tracing::debug;
 
 use crate::analytics::Analytics;
 use crate::extractors::authentication::policies::*;
@@ -22,6 +22,7 @@ mod api_key;
 mod dump;
 pub mod features;
 pub mod indexes;
+mod logs;
 mod metrics;
 mod multi_search;
 mod snapshot;
@@ -31,6 +32,7 @@ pub mod tasks;
 pub fn configure(cfg: &mut web::ServiceConfig) {
     cfg.service(web::scope("/tasks").configure(tasks::configure))
         .service(web::resource("/health").route(web::get().to(get_health)))
+        .service(web::scope("/logs").configure(logs::configure))
         .service(web::scope("/keys").configure(api_key::configure))
         .service(web::scope("/dumps").configure(dump::configure))
         .service(web::scope("/snapshots").configure(snapshot::configure))
@@ -250,7 +252,7 @@ async fn get_stats(
 
     let stats = create_all_stats((*index_scheduler).clone(), (*auth_controller).clone(), filters)?;
 
-    debug!("returns: {:?}", stats);
+    debug!(returns = ?stats, "Get stats");
     Ok(HttpResponse::Ok().json(stats))
 }
 

meilisearch/src/routes/multi_search.rs

@@ -3,16 +3,17 @@ use actix_web::web::{self, Data};
 use actix_web::{HttpRequest, HttpResponse};
 use deserr::actix_web::AwebJson;
 use index_scheduler::IndexScheduler;
-use log::debug;
 use meilisearch_types::deserr::DeserrJsonError;
 use meilisearch_types::error::ResponseError;
 use meilisearch_types::keys::actions;
 use serde::Serialize;
+use tracing::debug;
 
 use crate::analytics::{Analytics, MultiSearchAggregator};
 use crate::extractors::authentication::policies::ActionPolicy;
 use crate::extractors::authentication::{AuthenticationError, GuardedData};
 use crate::extractors::sequential_extractor::SeqHandler;
+use crate::routes::indexes::search::embed;
 use crate::search::{
     add_search_rules, perform_search, SearchQueryWithIndex, SearchResultWithIndex,
 };
@@ -51,7 +52,7 @@ pub async fn multi_search_with_post(
         for (query_index, (index_uid, mut query)) in
             queries.into_iter().map(SearchQueryWithIndex::into_index_query).enumerate()
         {
-            debug!("multi-search #{query_index}: called with params: {:?}", query);
+            debug!(on_index = query_index, parameters = ?query, "Multi-search");
 
             // Check index from API key
             if !index_scheduler.filters().is_index_authorized(&index_uid) {
@@ -74,10 +75,15 @@ pub async fn multi_search_with_post(
                 })
                 .with_index(query_index)?;
 
-            let search_result =
-                tokio::task::spawn_blocking(move || perform_search(&index, query, features))
-                    .await
-                    .with_index(query_index)?;
+            let distribution = embed(&mut query, index_scheduler.get_ref(), &index)
+                .await
+                .with_index(query_index)?;
+
+            let search_result = tokio::task::spawn_blocking(move || {
+                perform_search(&index, query, features, distribution)
+            })
+            .await
+            .with_index(query_index)?;
 
             search_results.push(SearchResultWithIndex {
                 index_uid: index_uid.into_inner(),
@@ -101,7 +107,7 @@ pub async fn multi_search_with_post(
         err
     })?;
 
-    debug!("returns: {:?}", search_results);
+    debug!(returns = ?search_results, "Multi-search");
 
     Ok(HttpResponse::Ok().json(SearchResults { results: search_results }))
 }

meilisearch/src/routes/snapshot.rs

@@ -1,10 +1,10 @@
 use actix_web::web::Data;
 use actix_web::{web, HttpRequest, HttpResponse};
 use index_scheduler::IndexScheduler;
-use log::debug;
 use meilisearch_types::error::ResponseError;
 use meilisearch_types::tasks::KindWithContent;
 use serde_json::json;
+use tracing::debug;
 
 use crate::analytics::Analytics;
 use crate::extractors::authentication::policies::*;
@@ -27,6 +27,6 @@ pub async fn create_snapshot(
     let task: SummarizedTaskView =
         tokio::task::spawn_blocking(move || index_scheduler.register(task)).await??.into();
 
-    debug!("returns: {:?}", task);
+    debug!(returns = ?task, "Create snapshot");
     Ok(HttpResponse::Accepted().json(task))
 }

meilisearch/src/routes/tasks.rs

@@ -8,11 +8,9 @@ use meilisearch_types::deserr::DeserrQueryParamError;
 use meilisearch_types::error::deserr_codes::*;
 use meilisearch_types::error::{InvalidTaskDateError, ResponseError};
 use meilisearch_types::index_uid::IndexUid;
-use meilisearch_types::settings::{Settings, Unchecked};
 use meilisearch_types::star_or::{OptionStarOr, OptionStarOrList};
-use meilisearch_types::tasks::{
-    serialize_duration, Details, IndexSwap, Kind, KindWithContent, Status, Task,
-};
+use meilisearch_types::task_view::TaskView;
+use meilisearch_types::tasks::{Kind, KindWithContent, Status};
 use serde::Serialize;
 use serde_json::json;
 use time::format_description::well_known::Rfc3339;
@@ -37,140 +35,6 @@ pub fn configure(cfg: &mut web::ServiceConfig) {
     .service(web::resource("/cancel").route(web::post().to(SeqHandler(cancel_tasks))))
     .service(web::resource("/{task_id}").route(web::get().to(SeqHandler(get_task))));
 }
-
-#[derive(Debug, Clone, PartialEq, Eq, Serialize)]
-#[serde(rename_all = "camelCase")]
-pub struct TaskView {
-    pub uid: TaskId,
-    #[serde(default)]
-    pub index_uid: Option<String>,
-    pub status: Status,
-    #[serde(rename = "type")]
-    pub kind: Kind,
-    pub canceled_by: Option<TaskId>,
-    #[serde(skip_serializing_if = "Option::is_none")]
-    pub details: Option<DetailsView>,
-    pub error: Option<ResponseError>,
-    #[serde(serialize_with = "serialize_duration", default)]
-    pub duration: Option<Duration>,
-    #[serde(with = "time::serde::rfc3339")]
-    pub enqueued_at: OffsetDateTime,
-    #[serde(with = "time::serde::rfc3339::option", default)]
-    pub started_at: Option<OffsetDateTime>,
-    #[serde(with = "time::serde::rfc3339::option", default)]
-    pub finished_at: Option<OffsetDateTime>,
-}
-
-impl TaskView {
-    pub fn from_task(task: &Task) -> TaskView {
-        TaskView {
-            uid: task.uid,
-            index_uid: task.index_uid().map(ToOwned::to_owned),
-            status: task.status,
-            kind: task.kind.as_kind(),
-            canceled_by: task.canceled_by,
-            details: task.details.clone().map(DetailsView::from),
-            error: task.error.clone(),
-            duration: task.started_at.zip(task.finished_at).map(|(start, end)| end - start),
-            enqueued_at: task.enqueued_at,
-            started_at: task.started_at,
-            finished_at: task.finished_at,
-        }
-    }
-}
-
-#[derive(Default, Debug, PartialEq, Eq, Clone, Serialize)]
-#[serde(rename_all = "camelCase")]
-pub struct DetailsView {
-    #[serde(skip_serializing_if = "Option::is_none")]
-    pub received_documents: Option<u64>,
-    #[serde(skip_serializing_if = "Option::is_none")]
-    pub indexed_documents: Option<Option<u64>>,
-    #[serde(skip_serializing_if = "Option::is_none")]
-    pub primary_key: Option<Option<String>>,
-    #[serde(skip_serializing_if = "Option::is_none")]
-    pub provided_ids: Option<usize>,
-    #[serde(skip_serializing_if = "Option::is_none")]
-    pub deleted_documents: Option<Option<u64>>,
-    #[serde(skip_serializing_if = "Option::is_none")]
-    pub matched_tasks: Option<u64>,
-    #[serde(skip_serializing_if = "Option::is_none")]
-    pub canceled_tasks: Option<Option<u64>>,
-    #[serde(skip_serializing_if = "Option::is_none")]
-    pub deleted_tasks: Option<Option<u64>>,
-    #[serde(skip_serializing_if = "Option::is_none")]
-    pub original_filter: Option<Option<String>>,
-    #[serde(skip_serializing_if = "Option::is_none")]
-    pub dump_uid: Option<Option<String>>,
-    #[serde(skip_serializing_if = "Option::is_none")]
-    #[serde(flatten)]
-    pub settings: Option<Box<Settings<Unchecked>>>,
-    #[serde(skip_serializing_if = "Option::is_none")]
-    pub swaps: Option<Vec<IndexSwap>>,
-}
-
-impl From<Details> for DetailsView {
-    fn from(details: Details) -> Self {
-        match details {
-            Details::DocumentAdditionOrUpdate { received_documents, indexed_documents } => {
-                DetailsView {
-                    received_documents: Some(received_documents),
-                    indexed_documents: Some(indexed_documents),
-                    ..DetailsView::default()
-                }
-            }
-            Details::SettingsUpdate { settings } => {
-                DetailsView { settings: Some(settings), ..DetailsView::default() }
-            }
-            Details::IndexInfo { primary_key } => {
-                DetailsView { primary_key: Some(primary_key), ..DetailsView::default() }
-            }
-            Details::DocumentDeletion {
-                provided_ids: received_document_ids,
-                deleted_documents,
-            } => DetailsView {
-                provided_ids: Some(received_document_ids),
-                deleted_documents: Some(deleted_documents),
-                original_filter: Some(None),
-                ..DetailsView::default()
-            },
-            Details::DocumentDeletionByFilter { original_filter, deleted_documents } => {
-                DetailsView {
-                    provided_ids: Some(0),
-                    original_filter: Some(Some(original_filter)),
-                    deleted_documents: Some(deleted_documents),
-                    ..DetailsView::default()
-                }
-            }
-            Details::ClearAll { deleted_documents } => {
-                DetailsView { deleted_documents: Some(deleted_documents), ..DetailsView::default() }
-            }
-            Details::TaskCancelation { matched_tasks, canceled_tasks, original_filter } => {
-                DetailsView {
-                    matched_tasks: Some(matched_tasks),
-                    canceled_tasks: Some(canceled_tasks),
-                    original_filter: Some(Some(original_filter)),
-                    ..DetailsView::default()
-                }
-            }
-            Details::TaskDeletion { matched_tasks, deleted_tasks, original_filter } => {
-                DetailsView {
-                    matched_tasks: Some(matched_tasks),
-                    deleted_tasks: Some(deleted_tasks),
-                    original_filter: Some(Some(original_filter)),
-                    ..DetailsView::default()
-                }
-            }
-            Details::Dump { dump_uid } => {
-                DetailsView { dump_uid: Some(dump_uid), ..DetailsView::default() }
-            }
-            Details::IndexSwap { swaps } => {
-                DetailsView { swaps: Some(swaps), ..Default::default() }
-            }
-        }
-    }
-}
-
 #[derive(Debug, Deserr)]
 #[deserr(error = DeserrQueryParamError, rename_all = camelCase, deny_unknown_fields)]
 pub struct TasksFilterQuery {

meilisearch/src/search.rs

@@ -7,24 +7,21 @@ use deserr::Deserr;
 use either::Either;
 use index_scheduler::RoFeatures;
 use indexmap::IndexMap;
-use log::warn;
 use meilisearch_auth::IndexSearchRules;
 use meilisearch_types::deserr::DeserrJsonError;
 use meilisearch_types::error::deserr_codes::*;
 use meilisearch_types::heed::RoTxn;
 use meilisearch_types::index_uid::IndexUid;
-use meilisearch_types::milli::score_details::{ScoreDetails, ScoringStrategy};
-use meilisearch_types::milli::{
-    dot_product_similarity, FacetValueHit, InternalError, OrderBy, SearchForFacetValues,
-};
+use meilisearch_types::milli::score_details::{self, ScoreDetails, ScoringStrategy};
+use meilisearch_types::milli::vector::DistributionShift;
+use meilisearch_types::milli::{FacetValueHit, OrderBy, SearchForFacetValues};
 use meilisearch_types::settings::DEFAULT_PAGINATION_MAX_TOTAL_HITS;
 use meilisearch_types::{milli, Document};
 use milli::tokenizer::TokenizerBuilder;
 use milli::{
     AscDesc, FieldId, FieldsIdsMap, Filter, FormatOptions, Index, MatchBounds, MatcherBuilder,
-    SortError, TermsMatchingStrategy, VectorOrArrayOfVectors, DEFAULT_VALUES_PER_FACET,
+    SortError, TermsMatchingStrategy, DEFAULT_VALUES_PER_FACET,
 };
-use ordered_float::OrderedFloat;
 use regex::Regex;
 use serde::Serialize;
 use serde_json::{json, Value};
@@ -39,6 +36,7 @@ pub const DEFAULT_CROP_LENGTH: fn() -> usize = || 10;
 pub const DEFAULT_CROP_MARKER: fn() -> String = || "…".to_string();
 pub const DEFAULT_HIGHLIGHT_PRE_TAG: fn() -> String = || "<em>".to_string();
 pub const DEFAULT_HIGHLIGHT_POST_TAG: fn() -> String = || "</em>".to_string();
+pub const DEFAULT_SEMANTIC_RATIO: fn() -> SemanticRatio = || SemanticRatio(0.5);
 
 #[derive(Debug, Clone, Default, PartialEq, Deserr)]
 #[deserr(error = DeserrJsonError, rename_all = camelCase, deny_unknown_fields)]
@@ -47,6 +45,8 @@ pub struct SearchQuery {
     pub q: Option<String>,
     #[deserr(default, error = DeserrJsonError<InvalidSearchVector>)]
     pub vector: Option<Vec<f32>>,
+    #[deserr(default, error = DeserrJsonError<InvalidHybridQuery>)]
+    pub hybrid: Option<HybridQuery>,
     #[deserr(default = DEFAULT_SEARCH_OFFSET(), error = DeserrJsonError<InvalidSearchOffset>)]
     pub offset: usize,
     #[deserr(default = DEFAULT_SEARCH_LIMIT(), error = DeserrJsonError<InvalidSearchLimit>)]
@@ -87,6 +87,48 @@ pub struct SearchQuery {
     pub attributes_to_search_on: Option<Vec<String>>,
 }
 
+#[derive(Debug, Clone, Default, PartialEq, Deserr)]
+#[deserr(error = DeserrJsonError<InvalidHybridQuery>, rename_all = camelCase, deny_unknown_fields)]
+pub struct HybridQuery {
+    /// TODO validate that sementic ratio is between 0.0 and 1,0
+    #[deserr(default, error = DeserrJsonError<InvalidSearchSemanticRatio>, default)]
+    pub semantic_ratio: SemanticRatio,
+    #[deserr(default, error = DeserrJsonError<InvalidEmbedder>, default)]
+    pub embedder: Option<String>,
+}
+
+#[derive(Debug, Clone, Copy, PartialEq, Deserr)]
+#[deserr(try_from(f32) = TryFrom::try_from -> InvalidSearchSemanticRatio)]
+pub struct SemanticRatio(f32);
+
+impl Default for SemanticRatio {
+    fn default() -> Self {
+        DEFAULT_SEMANTIC_RATIO()
+    }
+}
+
+impl std::convert::TryFrom<f32> for SemanticRatio {
+    type Error = InvalidSearchSemanticRatio;
+
+    fn try_from(f: f32) -> Result<Self, Self::Error> {
+        // the suggested "fix" is: `!(0.0..=1.0).contains(&f)`` which is allegedly less readable
+        #[allow(clippy::manual_range_contains)]
+        if f > 1.0 || f < 0.0 {
+            Err(InvalidSearchSemanticRatio)
+        } else {
+            Ok(SemanticRatio(f))
+        }
+    }
+}
+
+impl std::ops::Deref for SemanticRatio {
+    type Target = f32;
+
+    fn deref(&self) -> &Self::Target {
+        &self.0
+    }
+}
+
 impl SearchQuery {
     pub fn is_finite_pagination(&self) -> bool {
         self.page.or(self.hits_per_page).is_some()
@@ -106,6 +148,8 @@ pub struct SearchQueryWithIndex {
     pub q: Option<String>,
     #[deserr(default, error = DeserrJsonError<InvalidSearchQ>)]
     pub vector: Option<Vec<f32>>,
+    #[deserr(default, error = DeserrJsonError<InvalidHybridQuery>)]
+    pub hybrid: Option<HybridQuery>,
     #[deserr(default = DEFAULT_SEARCH_OFFSET(), error = DeserrJsonError<InvalidSearchOffset>)]
     pub offset: usize,
     #[deserr(default = DEFAULT_SEARCH_LIMIT(), error = DeserrJsonError<InvalidSearchLimit>)]
@@ -171,6 +215,7 @@ impl SearchQueryWithIndex {
             crop_marker,
             matching_strategy,
             attributes_to_search_on,
+            hybrid,
         } = self;
         (
             index_uid,
@@ -196,6 +241,7 @@ impl SearchQueryWithIndex {
                 crop_marker,
                 matching_strategy,
                 attributes_to_search_on,
+                hybrid,
                 // do not use ..Default::default() here,
                 // rather add any missing field from `SearchQuery` to `SearchQueryWithIndex`
             },
@@ -335,19 +381,44 @@ fn prepare_search<'t>(
     rtxn: &'t RoTxn,
     query: &'t SearchQuery,
     features: RoFeatures,
+    distribution: Option<DistributionShift>,
 ) -> Result<(milli::Search<'t>, bool, usize, usize), MeilisearchHttpError> {
     let mut search = index.search(rtxn);
 
-    if query.vector.is_some() && query.q.is_some() {
-        warn!("Ignoring the query string `q` when used with the `vector` parameter.");
+    if query.vector.is_some() {
+        features.check_vector("Passing `vector` as a query parameter")?;
     }
 
+    if query.hybrid.is_some() {
+        features.check_vector("Passing `hybrid` as a query parameter")?;
+    }
+
+    if query.hybrid.is_none() && query.q.is_some() && query.vector.is_some() {
+        return Err(MeilisearchHttpError::MissingSearchHybrid);
+    }
+
+    search.distribution_shift(distribution);
+
     if let Some(ref vector) = query.vector {
-        search.vector(vector.clone());
+        match &query.hybrid {
+            // If semantic ratio is 0.0, only the query search will impact the search results,
+            // skip the vector
+            Some(hybrid) if *hybrid.semantic_ratio == 0.0 => (),
+            _otherwise => {
+                search.vector(vector.clone());
+            }
+        }
     }
 
-    if let Some(ref query) = query.q {
-        search.query(query);
+    if let Some(ref q) = query.q {
+        match &query.hybrid {
+            // If semantic ratio is 1.0, only the vector search will impact the search results,
+            // skip the query
+            Some(hybrid) if *hybrid.semantic_ratio == 1.0 => (),
+            _otherwise => {
+                search.query(q);
+            }
+        }
     }
 
     if let Some(ref searchable) = query.attributes_to_search_on {
@@ -370,12 +441,8 @@ fn prepare_search<'t>(
         ScoringStrategy::Skip
     });
 
-    if query.show_ranking_score_details {
-        features.check_score_details()?;
-    }
-
-    if query.vector.is_some() {
-        features.check_vector()?;
+    if let Some(HybridQuery { embedder: Some(embedder), .. }) = &query.hybrid {
+        search.embedder_name(embedder);
     }
 
     // compute the offset on the limit depending on the pagination mode.
@@ -421,15 +488,22 @@ pub fn perform_search(
     index: &Index,
     query: SearchQuery,
     features: RoFeatures,
+    distribution: Option<DistributionShift>,
 ) -> Result<SearchResult, MeilisearchHttpError> {
     let before_search = Instant::now();
     let rtxn = index.read_txn()?;
 
     let (search, is_finite_pagination, max_total_hits, offset) =
-        prepare_search(index, &rtxn, &query, features)?;
+        prepare_search(index, &rtxn, &query, features, distribution)?;
 
     let milli::SearchResult { documents_ids, matching_words, candidates, document_scores, .. } =
-        search.execute()?;
+        match &query.hybrid {
+            Some(hybrid) => match *hybrid.semantic_ratio {
+                ratio if ratio == 0.0 || ratio == 1.0 => search.execute()?,
+                ratio => search.execute_hybrid(ratio)?,
+            },
+            None => search.execute()?,
+        };
 
     let fields_ids_map = index.fields_ids_map(&rtxn).unwrap();
 
@@ -538,13 +612,17 @@ pub fn perform_search(
             insert_geo_distance(sort, &mut document);
         }
 
-        let semantic_score = match query.vector.as_ref() {
-            Some(vector) => match extract_field("_vectors", &fields_ids_map, obkv)? {
-                Some(vectors) => compute_semantic_score(vector, vectors)?,
-                None => None,
-            },
-            None => None,
-        };
+        let mut semantic_score = None;
+        for details in &score {
+            if let ScoreDetails::Vector(score_details::Vector {
+                target_vector: _,
+                value_similarity: Some((_matching_vector, similarity)),
+            }) = details
+            {
+                semantic_score = Some(*similarity);
+                break;
+            }
+        }
 
         let ranking_score =
             query.show_ranking_score.then(|| ScoreDetails::global_score(score.iter()));
@@ -647,11 +725,15 @@ pub fn perform_facet_search(
     let before_search = Instant::now();
     let rtxn = index.read_txn()?;
 
-    let (search, _, _, _) = prepare_search(index, &rtxn, &search_query, features)?;
-    let mut facet_search = SearchForFacetValues::new(facet_name, search);
+    let (search, _, _, _) = prepare_search(index, &rtxn, &search_query, features, None)?;
+    let mut facet_search =
+        SearchForFacetValues::new(facet_name, search, search_query.hybrid.is_some());
     if let Some(facet_query) = &facet_query {
         facet_search.query(facet_query);
     }
+    if let Some(max_facets) = index.max_values_per_facet(&rtxn)? {
+        facet_search.max_values(max_facets as usize);
+    }
 
     Ok(FacetSearchResult {
         facet_hits: facet_search.execute()?,
@@ -676,18 +758,6 @@ fn insert_geo_distance(sorts: &[String], document: &mut Document) {
     }
 }
 
-fn compute_semantic_score(query: &[f32], vectors: Value) -> milli::Result<Option<f32>> {
-    let vectors = serde_json::from_value(vectors)
-        .map(VectorOrArrayOfVectors::into_array_of_vectors)
-        .map_err(InternalError::SerdeJson)?;
-    Ok(vectors
-        .into_iter()
-        .flatten()
-        .map(|v| OrderedFloat(dot_product_similarity(query, &v)))
-        .max()
-        .map(OrderedFloat::into_inner))
-}
-
 fn compute_formatted_options(
     attr_to_highlight: &HashSet<String>,
     attr_to_crop: &[String],
@@ -815,22 +885,6 @@ fn make_document(
     Ok(document)
 }
 
-/// Extract the JSON value under the field name specified
-/// but doesn't support nested objects.
-fn extract_field(
-    field_name: &str,
-    field_ids_map: &FieldsIdsMap,
-    obkv: obkv::KvReaderU16,
-) -> Result<Option<serde_json::Value>, MeilisearchHttpError> {
-    match field_ids_map.id(field_name) {
-        Some(fid) => match obkv.get(fid) {
-            Some(value) => Ok(serde_json::from_slice(value).map(Some)?),
-            None => Ok(None),
-        },
-        None => Ok(None),
-    }
-}
-
 fn format_fields<'a>(
     document: &Document,
     field_ids_map: &FieldsIdsMap,
@@ -842,6 +896,14 @@ fn format_fields<'a>(
     let mut matches_position = compute_matches.then(BTreeMap::new);
     let mut document = document.clone();
 
+    // reduce the formatted option list to the attributes that should be formatted,
+    // instead of all the attributes to display.
+    let formatting_fields_options: Vec<_> = formatted_options
+        .iter()
+        .filter(|(_, option)| option.should_format())
+        .map(|(fid, option)| (field_ids_map.name(*fid).unwrap(), option))
+        .collect();
+
     // select the attributes to retrieve
     let displayable_names =
         displayable_ids.iter().map(|&fid| field_ids_map.name(fid).expect("Missing field name"));
@@ -850,13 +912,15 @@ fn format_fields<'a>(
         // to the value and merge them together. eg. If a user said he wanted to highlight `doggo`
         // and crop `doggo.name`. `doggo.name` needs to be highlighted + cropped while `doggo.age` is only
         // highlighted.
-        let format = formatted_options
+        // Warn: The time to compute the format list scales with the number of fields to format;
+        // cumulated with map_leaf_values that iterates over all the nested fields, it gives a quadratic complexity:
+        // d*f where d is the total number of fields to display and f is the total number of fields to format.
+        let format = formatting_fields_options
             .iter()
-            .filter(|(field, _option)| {
-                let name = field_ids_map.name(**field).unwrap();
+            .filter(|(name, _option)| {
                 milli::is_faceted_by(name, key) || milli::is_faceted_by(key, name)
             })
-            .map(|(_, option)| *option)
+            .map(|(_, option)| **option)
             .reduce(|acc, option| acc.merge(option));
         let mut infos = Vec::new();
 
@@ -953,7 +1017,7 @@ fn format_value<'a>(
                     let value = matcher.format(format_options);
                     Value::String(value.into_owned())
                 }
-                None => Value::Number(number),
+                None => Value::String(s),
             }
         }
         value => value,

meilisearch/tests/auth/authorization.rs

@@ -59,6 +59,8 @@ pub static AUTHORIZATIONS: Lazy<HashMap<(&'static str, &'static str), HashSet<&'
             ("POST",    "/snapshots") =>                                       hashset!{"snapshots.create", "snapshots.*", "*"},
             ("GET",     "/version") =>                                         hashset!{"version", "*"},
             ("GET",     "/metrics") =>                                         hashset!{"metrics.get", "metrics.*", "*"},
+            ("POST",    "/logs/stream") =>                                     hashset!{"metrics.get", "metrics.*", "*"},
+            ("DELETE",  "/logs/stream") =>                                     hashset!{"metrics.get", "metrics.*", "*"},
             ("PATCH",   "/keys/mykey/") =>                                     hashset!{"keys.update", "*"},
             ("GET",     "/keys/mykey/") =>                                     hashset!{"keys.get", "*"},
             ("DELETE",  "/keys/mykey/") =>                                     hashset!{"keys.delete", "*"},

meilisearch/tests/common/mod.rs

@@ -64,7 +64,7 @@ impl Display for Value {
         write!(
             f,
             "{}",
-            json_string!(self, { ".enqueuedAt" => "[date]", ".processedAt" => "[date]", ".finishedAt" => "[date]", ".duration" => "[duration]" })
+            json_string!(self, { ".enqueuedAt" => "[date]", ".startedAt" => "[date]", ".finishedAt" => "[date]", ".duration" => "[duration]" })
         )
     }
 }

meilisearch/tests/common/server.rs

@@ -13,6 +13,8 @@ use meilisearch::{analytics, create_app, setup_meilisearch};
 use once_cell::sync::Lazy;
 use tempfile::TempDir;
 use tokio::time::sleep;
+use tracing::level_filters::LevelFilter;
+use tracing_subscriber::Layer;
 
 use super::index::Index;
 use super::service::Service;
@@ -81,10 +83,16 @@ impl Server {
         Response = ServiceResponse<impl MessageBody>,
         Error = actix_web::Error,
     > {
+        let (_route_layer, route_layer_handle) =
+            tracing_subscriber::reload::Layer::new(None.with_filter(
+                tracing_subscriber::filter::Targets::new().with_target("", LevelFilter::OFF),
+            ));
+
         actix_web::test::init_service(create_app(
             self.service.index_scheduler.clone().into(),
             self.service.auth.clone().into(),
             self.service.options.clone(),
+            route_layer_handle,
             analytics::MockAnalytics::new(&self.service.options),
             true,
         ))

meilisearch/tests/common/service.rs

@@ -7,6 +7,8 @@ use actix_web::test::TestRequest;
 use index_scheduler::IndexScheduler;
 use meilisearch::{analytics, create_app, Opt};
 use meilisearch_auth::AuthController;
+use tracing::level_filters::LevelFilter;
+use tracing_subscriber::Layer;
 
 use crate::common::encoder::Encoder;
 use crate::common::Value;
@@ -105,10 +107,16 @@ impl Service {
     }
 
     pub async fn request(&self, mut req: test::TestRequest) -> (Value, StatusCode) {
+        let (_route_layer, route_layer_handle) =
+            tracing_subscriber::reload::Layer::new(None.with_filter(
+                tracing_subscriber::filter::Targets::new().with_target("", LevelFilter::OFF),
+            ));
+
         let app = test::init_service(create_app(
             self.index_scheduler.clone().into(),
             self.auth.clone().into(),
             self.options.clone(),
+            route_layer_handle,
             analytics::MockAnalytics::new(&self.options),
             true,
         ))

meilisearch/tests/documents/add_documents.rs

@@ -1760,6 +1760,181 @@ async fn add_documents_invalid_geo_field() {
       "finishedAt": "[date]"
     }
     "###);
+
+    // The three next tests are related to #4333
+
+    // _geo has a lat and lng but set to `null`
+    let documents = json!([
+        {
+            "id": "12",
+            "_geo": { "lng": null, "lat": 67}
+        }
+    ]);
+
+    let (response, code) = index.add_documents(documents, None).await;
+    snapshot!(code, @"202 Accepted");
+    let response = index.wait_task(response.uid()).await;
+    snapshot!(json_string!(response, { ".duration" => "[duration]", ".enqueuedAt" => "[date]", ".startedAt" => "[date]", ".finishedAt" => "[date]" }),
+        @r###"
+    {
+      "uid": 14,
+      "indexUid": "test",
+      "status": "failed",
+      "type": "documentAdditionOrUpdate",
+      "canceledBy": null,
+      "details": {
+        "receivedDocuments": 1,
+        "indexedDocuments": 0
+      },
+      "error": {
+        "message": "Could not parse longitude in the document with the id: `12`. Was expecting a finite number but instead got `null`.",
+        "code": "invalid_document_geo_field",
+        "type": "invalid_request",
+        "link": "https://docs.meilisearch.com/errors#invalid_document_geo_field"
+      },
+      "duration": "[duration]",
+      "enqueuedAt": "[date]",
+      "startedAt": "[date]",
+      "finishedAt": "[date]"
+    }
+    "###);
+
+    // _geo has a lat and lng but set to `null`
+    let documents = json!([
+        {
+            "id": "12",
+            "_geo": { "lng": 35, "lat": null }
+        }
+    ]);
+
+    let (response, code) = index.add_documents(documents, None).await;
+    snapshot!(code, @"202 Accepted");
+    let response = index.wait_task(response.uid()).await;
+    snapshot!(json_string!(response, { ".duration" => "[duration]", ".enqueuedAt" => "[date]", ".startedAt" => "[date]", ".finishedAt" => "[date]" }),
+        @r###"
+    {
+      "uid": 15,
+      "indexUid": "test",
+      "status": "failed",
+      "type": "documentAdditionOrUpdate",
+      "canceledBy": null,
+      "details": {
+        "receivedDocuments": 1,
+        "indexedDocuments": 0
+      },
+      "error": {
+        "message": "Could not parse latitude in the document with the id: `12`. Was expecting a finite number but instead got `null`.",
+        "code": "invalid_document_geo_field",
+        "type": "invalid_request",
+        "link": "https://docs.meilisearch.com/errors#invalid_document_geo_field"
+      },
+      "duration": "[duration]",
+      "enqueuedAt": "[date]",
+      "startedAt": "[date]",
+      "finishedAt": "[date]"
+    }
+    "###);
+
+    // _geo has a lat and lng but set to `null`
+    let documents = json!([
+        {
+            "id": "13",
+            "_geo": { "lng": null, "lat": null }
+        }
+    ]);
+
+    let (response, code) = index.add_documents(documents, None).await;
+    snapshot!(code, @"202 Accepted");
+    let response = index.wait_task(response.uid()).await;
+    snapshot!(json_string!(response, { ".duration" => "[duration]", ".enqueuedAt" => "[date]", ".startedAt" => "[date]", ".finishedAt" => "[date]" }),
+        @r###"
+    {
+      "uid": 16,
+      "indexUid": "test",
+      "status": "failed",
+      "type": "documentAdditionOrUpdate",
+      "canceledBy": null,
+      "details": {
+        "receivedDocuments": 1,
+        "indexedDocuments": 0
+      },
+      "error": {
+        "message": "Could not parse latitude nor longitude in the document with the id: `13`. Was expecting finite numbers but instead got `null` and `null`.",
+        "code": "invalid_document_geo_field",
+        "type": "invalid_request",
+        "link": "https://docs.meilisearch.com/errors#invalid_document_geo_field"
+      },
+      "duration": "[duration]",
+      "enqueuedAt": "[date]",
+      "startedAt": "[date]",
+      "finishedAt": "[date]"
+    }
+    "###);
+}
+
+// Related to #4333
+#[actix_rt::test]
+async fn add_invalid_geo_and_then_settings() {
+    let server = Server::new().await;
+    let index = server.index("test");
+    index.create(Some("id")).await;
+
+    // _geo is not an object
+    let documents = json!([
+        {
+            "id": "11",
+            "_geo": { "lat": null, "lng": null },
+        }
+    ]);
+    let (ret, code) = index.add_documents(documents, None).await;
+    snapshot!(code, @"202 Accepted");
+    let ret = index.wait_task(ret.uid()).await;
+    snapshot!(ret, @r###"
+    {
+      "uid": 1,
+      "indexUid": "test",
+      "status": "succeeded",
+      "type": "documentAdditionOrUpdate",
+      "canceledBy": null,
+      "details": {
+        "receivedDocuments": 1,
+        "indexedDocuments": 1
+      },
+      "error": null,
+      "duration": "[duration]",
+      "enqueuedAt": "[date]",
+      "startedAt": "[date]",
+      "finishedAt": "[date]"
+    }
+    "###);
+
+    let (ret, code) = index.update_settings(json!({"sortableAttributes": ["_geo"]})).await;
+    snapshot!(code, @"202 Accepted");
+    let ret = index.wait_task(ret.uid()).await;
+    snapshot!(ret, @r###"
+    {
+      "uid": 2,
+      "indexUid": "test",
+      "status": "failed",
+      "type": "settingsUpdate",
+      "canceledBy": null,
+      "details": {
+        "sortableAttributes": [
+          "_geo"
+        ]
+      },
+      "error": {
+        "message": "Could not parse latitude in the document with the id: `\"11\"`. Was expecting a finite number but instead got `null`.",
+        "code": "invalid_document_geo_field",
+        "type": "invalid_request",
+        "link": "https://docs.meilisearch.com/errors#invalid_document_geo_field"
+      },
+      "duration": "[duration]",
+      "enqueuedAt": "[date]",
+      "startedAt": "[date]",
+      "finishedAt": "[date]"
+    }
+    "###);
 }
 
 #[actix_rt::test]

meilisearch/tests/dumps/mod.rs

@@ -59,7 +59,7 @@ async fn import_dump_v1_movie_raw() {
       "dictionary": [],
       "synonyms": {},
       "distinctAttribute": null,
-      "proximityPrecision": null,
+      "proximityPrecision": "byWord",
       "typoTolerance": {
         "enabled": true,
         "minWordSizeForTypos": {
@@ -220,7 +220,7 @@ async fn import_dump_v1_movie_with_settings() {
       "dictionary": [],
       "synonyms": {},
       "distinctAttribute": null,
-      "proximityPrecision": null,
+      "proximityPrecision": "byWord",
       "typoTolerance": {
         "enabled": true,
         "minWordSizeForTypos": {
@@ -367,7 +367,7 @@ async fn import_dump_v1_rubygems_with_settings() {
       "dictionary": [],
       "synonyms": {},
       "distinctAttribute": null,
-      "proximityPrecision": null,
+      "proximityPrecision": "byWord",
       "typoTolerance": {
         "enabled": true,
         "minWordSizeForTypos": {
@@ -500,7 +500,7 @@ async fn import_dump_v2_movie_raw() {
       "dictionary": [],
       "synonyms": {},
       "distinctAttribute": null,
-      "proximityPrecision": null,
+      "proximityPrecision": "byWord",
       "typoTolerance": {
         "enabled": true,
         "minWordSizeForTypos": {
@@ -645,7 +645,7 @@ async fn import_dump_v2_movie_with_settings() {
       "dictionary": [],
       "synonyms": {},
       "distinctAttribute": null,
-      "proximityPrecision": null,
+      "proximityPrecision": "byWord",
       "typoTolerance": {
         "enabled": true,
         "minWordSizeForTypos": {
@@ -789,7 +789,7 @@ async fn import_dump_v2_rubygems_with_settings() {
       "dictionary": [],
       "synonyms": {},
       "distinctAttribute": null,
-      "proximityPrecision": null,
+      "proximityPrecision": "byWord",
       "typoTolerance": {
         "enabled": true,
         "minWordSizeForTypos": {
@@ -922,7 +922,7 @@ async fn import_dump_v3_movie_raw() {
       "dictionary": [],
       "synonyms": {},
       "distinctAttribute": null,
-      "proximityPrecision": null,
+      "proximityPrecision": "byWord",
       "typoTolerance": {
         "enabled": true,
         "minWordSizeForTypos": {
@@ -1067,7 +1067,7 @@ async fn import_dump_v3_movie_with_settings() {
       "dictionary": [],
       "synonyms": {},
       "distinctAttribute": null,
-      "proximityPrecision": null,
+      "proximityPrecision": "byWord",
       "typoTolerance": {
         "enabled": true,
         "minWordSizeForTypos": {
@@ -1211,7 +1211,7 @@ async fn import_dump_v3_rubygems_with_settings() {
       "dictionary": [],
       "synonyms": {},
       "distinctAttribute": null,
-      "proximityPrecision": null,
+      "proximityPrecision": "byWord",
       "typoTolerance": {
         "enabled": true,
         "minWordSizeForTypos": {
@@ -1344,7 +1344,7 @@ async fn import_dump_v4_movie_raw() {
       "dictionary": [],
       "synonyms": {},
       "distinctAttribute": null,
-      "proximityPrecision": null,
+      "proximityPrecision": "byWord",
       "typoTolerance": {
         "enabled": true,
         "minWordSizeForTypos": {
@@ -1489,7 +1489,7 @@ async fn import_dump_v4_movie_with_settings() {
       "dictionary": [],
       "synonyms": {},
       "distinctAttribute": null,
-      "proximityPrecision": null,
+      "proximityPrecision": "byWord",
       "typoTolerance": {
         "enabled": true,
         "minWordSizeForTypos": {
@@ -1633,7 +1633,7 @@ async fn import_dump_v4_rubygems_with_settings() {
       "dictionary": [],
       "synonyms": {},
       "distinctAttribute": null,
-      "proximityPrecision": null,
+      "proximityPrecision": "byWord",
       "typoTolerance": {
         "enabled": true,
         "minWordSizeForTypos": {
@@ -1845,11 +1845,10 @@ async fn import_dump_v6_containing_experimental_features() {
     meili_snap::snapshot!(code, @"200 OK");
     meili_snap::snapshot!(meili_snap::json_string!(response), @r###"
     {
-      "scoreDetails": false,
       "vectorStore": false,
       "metrics": false,
-      "exportPuffinReports": false,
-      "proximityPrecision": false
+      "logsRoute": false,
+      "exportPuffinReports": false
     }
     "###);
 
@@ -1878,7 +1877,7 @@ async fn import_dump_v6_containing_experimental_features() {
       "dictionary": [],
       "synonyms": {},
       "distinctAttribute": null,
-      "proximityPrecision": "attributeScale",
+      "proximityPrecision": "byAttribute",
       "typoTolerance": {
         "enabled": true,
         "minWordSizeForTypos": {

meilisearch/tests/features/mod.rs

@@ -18,11 +18,10 @@ async fn experimental_features() {
     meili_snap::snapshot!(code, @"200 OK");
     meili_snap::snapshot!(meili_snap::json_string!(response), @r###"
     {
-      "scoreDetails": false,
       "vectorStore": false,
       "metrics": false,
-      "exportPuffinReports": false,
-      "proximityPrecision": false
+      "logsRoute": false,
+      "exportPuffinReports": false
     }
     "###);
 
@@ -31,11 +30,10 @@ async fn experimental_features() {
     meili_snap::snapshot!(code, @"200 OK");
     meili_snap::snapshot!(meili_snap::json_string!(response), @r###"
     {
-      "scoreDetails": false,
       "vectorStore": true,
       "metrics": false,
-      "exportPuffinReports": false,
-      "proximityPrecision": false
+      "logsRoute": false,
+      "exportPuffinReports": false
     }
     "###);
 
@@ -44,11 +42,10 @@ async fn experimental_features() {
     meili_snap::snapshot!(code, @"200 OK");
     meili_snap::snapshot!(meili_snap::json_string!(response), @r###"
     {
-      "scoreDetails": false,
       "vectorStore": true,
       "metrics": false,
-      "exportPuffinReports": false,
-      "proximityPrecision": false
+      "logsRoute": false,
+      "exportPuffinReports": false
     }
     "###);
 
@@ -58,11 +55,10 @@ async fn experimental_features() {
     meili_snap::snapshot!(code, @"200 OK");
     meili_snap::snapshot!(meili_snap::json_string!(response), @r###"
     {
-      "scoreDetails": false,
       "vectorStore": true,
       "metrics": false,
-      "exportPuffinReports": false,
-      "proximityPrecision": false
+      "logsRoute": false,
+      "exportPuffinReports": false
     }
     "###);
 
@@ -72,11 +68,10 @@ async fn experimental_features() {
     meili_snap::snapshot!(code, @"200 OK");
     meili_snap::snapshot!(meili_snap::json_string!(response), @r###"
     {
-      "scoreDetails": false,
       "vectorStore": true,
       "metrics": false,
-      "exportPuffinReports": false,
-      "proximityPrecision": false
+      "logsRoute": false,
+      "exportPuffinReports": false
     }
     "###);
 }
@@ -93,11 +88,10 @@ async fn experimental_feature_metrics() {
     meili_snap::snapshot!(code, @"200 OK");
     meili_snap::snapshot!(meili_snap::json_string!(response), @r###"
     {
-      "scoreDetails": false,
       "vectorStore": false,
       "metrics": true,
-      "exportPuffinReports": false,
-      "proximityPrecision": false
+      "logsRoute": false,
+      "exportPuffinReports": false
     }
     "###);
 
@@ -152,7 +146,7 @@ async fn errors() {
     meili_snap::snapshot!(code, @"400 Bad Request");
     meili_snap::snapshot!(meili_snap::json_string!(response), @r###"
     {
-      "message": "Unknown field `NotAFeature`: expected one of `scoreDetails`, `vectorStore`, `metrics`, `exportPuffinReports`, `proximityPrecision`",
+      "message": "Unknown field `NotAFeature`: expected one of `vectorStore`, `metrics`, `logsRoute`, `exportPuffinReports`",
       "code": "bad_request",
       "type": "invalid_request",
       "link": "https://docs.meilisearch.com/errors#bad_request"

meilisearch/tests/integration.rs

@@ -5,6 +5,7 @@ mod documents;
 mod dumps;
 mod features;
 mod index;
+mod logs;
 mod search;
 mod settings;
 mod snapshot;

meilisearch/tests/logs/error.rs

@@ -0,0 +1,182 @@
+use meili_snap::*;
+
+use crate::common::Server;
+use crate::json;
+
+#[actix_rt::test]
+async fn logs_stream_bad_target() {
+    let server = Server::new().await;
+
+    // Wrong type
+    let (response, code) = server.service.post("/logs/stream", json!({ "target": true })).await;
+    snapshot!(code, @"400 Bad Request");
+    snapshot!(response, @r###"
+    {
+      "message": "Invalid value type at `.target`: expected a string, but found a boolean: `true`",
+      "code": "bad_request",
+      "type": "invalid_request",
+      "link": "https://docs.meilisearch.com/errors#bad_request"
+    }
+    "###);
+
+    // Wrong type
+    let (response, code) = server.service.post("/logs/stream", json!({ "target": [] })).await;
+    snapshot!(code, @"400 Bad Request");
+    snapshot!(response, @r###"
+    {
+      "message": "Invalid value type at `.target`: expected a string, but found an array: `[]`",
+      "code": "bad_request",
+      "type": "invalid_request",
+      "link": "https://docs.meilisearch.com/errors#bad_request"
+    }
+    "###);
+
+    // Our help message
+    let (response, code) = server.service.post("/logs/stream", json!({ "target": "" })).await;
+    snapshot!(code, @"400 Bad Request");
+    snapshot!(response, @r###"
+    {
+      "message": "Invalid value at `.target`: Empty string is not a valid target. If you want to get no logs use `OFF`. Usage: `info`, `meilisearch=info`, or you can write multiple filters in one target: `index_scheduler=info,milli=trace`",
+      "code": "bad_request",
+      "type": "invalid_request",
+      "link": "https://docs.meilisearch.com/errors#bad_request"
+    }
+    "###);
+
+    // An error from the target parser
+    let (response, code) = server.service.post("/logs/stream", json!({ "target": "==" })).await;
+    snapshot!(code, @"400 Bad Request");
+    snapshot!(response, @r###"
+    {
+      "message": "Invalid value at `.target`: invalid filter directive: too many '=' in filter directive, expected 0 or 1",
+      "code": "bad_request",
+      "type": "invalid_request",
+      "link": "https://docs.meilisearch.com/errors#bad_request"
+    }
+    "###);
+}
+
+#[actix_rt::test]
+async fn logs_stream_bad_mode() {
+    let server = Server::new().await;
+
+    // Wrong type
+    let (response, code) = server.service.post("/logs/stream", json!({ "mode": true })).await;
+    snapshot!(code, @"400 Bad Request");
+    snapshot!(response, @r###"
+    {
+      "message": "Invalid value type at `.mode`: expected a string, but found a boolean: `true`",
+      "code": "bad_request",
+      "type": "invalid_request",
+      "link": "https://docs.meilisearch.com/errors#bad_request"
+    }
+    "###);
+
+    // Wrong type
+    let (response, code) = server.service.post("/logs/stream", json!({ "mode": [] })).await;
+    snapshot!(code, @"400 Bad Request");
+    snapshot!(response, @r###"
+    {
+      "message": "Invalid value type at `.mode`: expected a string, but found an array: `[]`",
+      "code": "bad_request",
+      "type": "invalid_request",
+      "link": "https://docs.meilisearch.com/errors#bad_request"
+    }
+    "###);
+
+    // Wrong value
+    let (response, code) = server.service.post("/logs/stream", json!({ "mode": "tamo" })).await;
+    snapshot!(code, @"400 Bad Request");
+    snapshot!(response, @r###"
+    {
+      "message": "Unknown value `tamo` at `.mode`: expected one of `human`, `profile`",
+      "code": "bad_request",
+      "type": "invalid_request",
+      "link": "https://docs.meilisearch.com/errors#bad_request"
+    }
+    "###);
+}
+
+#[actix_rt::test]
+async fn logs_stream_bad_profile_memory() {
+    let server = Server::new().await;
+
+    // Wrong type
+    let (response, code) =
+        server.service.post("/logs/stream", json!({ "profileMemory": "tamo" })).await;
+    snapshot!(code, @"400 Bad Request");
+    snapshot!(response, @r###"
+    {
+      "message": "Invalid value type at `.profileMemory`: expected a boolean, but found a string: `\"tamo\"`",
+      "code": "bad_request",
+      "type": "invalid_request",
+      "link": "https://docs.meilisearch.com/errors#bad_request"
+    }
+    "###);
+
+    // Wrong type
+    let (response, code) =
+        server.service.post("/logs/stream", json!({ "profileMemory": ["hello", "kefir"] })).await;
+    snapshot!(code, @"400 Bad Request");
+    snapshot!(response, @r###"
+    {
+      "message": "Invalid value type at `.profileMemory`: expected a boolean, but found an array: `[\"hello\",\"kefir\"]`",
+      "code": "bad_request",
+      "type": "invalid_request",
+      "link": "https://docs.meilisearch.com/errors#bad_request"
+    }
+    "###);
+
+    // Used with default parameters
+    let (response, code) =
+        server.service.post("/logs/stream", json!({ "profileMemory": true })).await;
+    snapshot!(code, @"400 Bad Request");
+    snapshot!(response, @r###"
+    {
+      "message": "Invalid value: `profile_memory` can only be used while profiling code and is not compatible with the Human mode.",
+      "code": "invalid_settings_typo_tolerance",
+      "type": "invalid_request",
+      "link": "https://docs.meilisearch.com/errors#invalid_settings_typo_tolerance"
+    }
+    "###);
+
+    // Used with an unsupported mode
+    let (response, code) =
+        server.service.post("/logs/stream", json!({ "mode": "fmt", "profileMemory": true })).await;
+    snapshot!(code, @"400 Bad Request");
+    snapshot!(response, @r###"
+    {
+      "message": "Unknown value `fmt` at `.mode`: expected one of `human`, `profile`",
+      "code": "bad_request",
+      "type": "invalid_request",
+      "link": "https://docs.meilisearch.com/errors#bad_request"
+    }
+    "###);
+}
+
+#[actix_rt::test]
+async fn logs_stream_without_enabling_the_route() {
+    let server = Server::new().await;
+
+    let (response, code) = server.service.post("/logs/stream", json!({})).await;
+    snapshot!(code, @"400 Bad Request");
+    snapshot!(response, @r###"
+    {
+      "message": "getting logs through the `/logs/stream` route requires enabling the `logs route` experimental feature. See https://github.com/orgs/meilisearch/discussions/721",
+      "code": "feature_not_enabled",
+      "type": "invalid_request",
+      "link": "https://docs.meilisearch.com/errors#feature_not_enabled"
+    }
+    "###);
+
+    let (response, code) = server.service.delete("/logs/stream").await;
+    snapshot!(code, @"400 Bad Request");
+    snapshot!(response, @r###"
+    {
+      "message": "getting logs through the `/logs/stream` route requires enabling the `logs route` experimental feature. See https://github.com/orgs/meilisearch/discussions/721",
+      "code": "feature_not_enabled",
+      "type": "invalid_request",
+      "link": "https://docs.meilisearch.com/errors#feature_not_enabled"
+    }
+    "###);
+}

meilisearch/tests/logs/mod.rs

@@ -0,0 +1,92 @@
+mod error;
+
+use std::rc::Rc;
+use std::str::FromStr;
+
+use actix_web::http::header::ContentType;
+use meili_snap::snapshot;
+use meilisearch::{analytics, create_app, Opt};
+use tracing::level_filters::LevelFilter;
+use tracing_subscriber::layer::SubscriberExt;
+use tracing_subscriber::Layer;
+
+use crate::common::{default_settings, Server};
+use crate::json;
+
+#[actix_web::test]
+async fn basic_test_log_stream_route() {
+    let db_path = tempfile::tempdir().unwrap();
+    let server = Server::new_with_options(Opt {
+        experimental_enable_logs_route: true,
+        ..default_settings(db_path.path())
+    })
+    .await
+    .unwrap();
+
+    let (route_layer, route_layer_handle) =
+        tracing_subscriber::reload::Layer::new(None.with_filter(
+            tracing_subscriber::filter::Targets::new().with_target("", LevelFilter::OFF),
+        ));
+
+    let subscriber = tracing_subscriber::registry().with(route_layer).with(
+        tracing_subscriber::fmt::layer()
+            .with_span_events(tracing_subscriber::fmt::format::FmtSpan::ACTIVE)
+            .with_filter(tracing_subscriber::filter::LevelFilter::from_str("INFO").unwrap()),
+    );
+
+    let app = actix_web::test::init_service(create_app(
+        server.service.index_scheduler.clone().into(),
+        server.service.auth.clone().into(),
+        server.service.options.clone(),
+        route_layer_handle,
+        analytics::MockAnalytics::new(&server.service.options),
+        true,
+    ))
+    .await;
+
+    // set the subscriber as the default for the application
+    tracing::subscriber::set_global_default(subscriber).unwrap();
+
+    let app = Rc::new(app);
+
+    // First, we start listening on the `/logs/stream` route
+    let handle_app = app.clone();
+    let handle = tokio::task::spawn_local(async move {
+        let req = actix_web::test::TestRequest::post()
+            .uri("/logs/stream")
+            .insert_header(ContentType::json())
+            .set_payload(
+                serde_json::to_vec(&json!({
+                    "mode": "human",
+                    "target": "info",
+                }))
+                .unwrap(),
+            );
+        let req = req.to_request();
+        let ret = actix_web::test::call_service(&*handle_app, req).await;
+        actix_web::test::read_body(ret).await
+    });
+
+    // We're going to create an index to get at least one info log saying we processed a batch of task
+    let (ret, _code) = server.create_index(json!({ "uid": "tamo" })).await;
+    snapshot!(ret, @r###"
+    {
+      "taskUid": 0,
+      "indexUid": "tamo",
+      "status": "enqueued",
+      "type": "indexCreation",
+      "enqueuedAt": "[date]"
+    }
+    "###);
+    server.wait_task(ret.uid()).await;
+
+    let req = actix_web::test::TestRequest::delete().uri("/logs/stream");
+    let req = req.to_request();
+    let ret = actix_web::test::call_service(&*app, req).await;
+    let code = ret.status();
+    snapshot!(code, @"204 No Content");
+
+    let logs = handle.await.unwrap();
+    let logs = String::from_utf8(logs.to_vec()).unwrap();
+    assert!(logs.contains("INFO"), "{logs}");
+}

meilisearch/tests/search/facet_search.rs

@@ -105,6 +105,24 @@ async fn more_advanced_facet_search() {
     snapshot!(response["facetHits"].as_array().unwrap().len(), @"1");
 }
 
+#[actix_rt::test]
+async fn simple_facet_search_with_max_values() {
+    let server = Server::new().await;
+    let index = server.index("test");
+
+    let documents = DOCUMENTS.clone();
+    index.update_settings_faceting(json!({ "maxValuesPerFacet": 1 })).await;
+    index.update_settings_filterable_attributes(json!(["genres"])).await;
+    index.add_documents(documents, None).await;
+    index.wait_task(2).await;
+
+    let (response, code) =
+        index.facet_search(json!({"facetName": "genres", "facetQuery": "a"})).await;
+
+    assert_eq!(code, 200, "{}", response);
+    assert_eq!(dbg!(response)["facetHits"].as_array().unwrap().len(), 1);
+}
+
 #[actix_rt::test]
 async fn non_filterable_facet_search_error() {
     let server = Server::new().await;

meilisearch/tests/search/hybrid.rs

@@ -0,0 +1,221 @@
+use meili_snap::snapshot;
+use once_cell::sync::Lazy;
+
+use crate::common::index::Index;
+use crate::common::{Server, Value};
+use crate::json;
+
+async fn index_with_documents<'a>(server: &'a Server, documents: &Value) -> Index<'a> {
+    let index = server.index("test");
+
+    let (response, code) = server.set_features(json!({"vectorStore": true})).await;
+
+    meili_snap::snapshot!(code, @"200 OK");
+    meili_snap::snapshot!(meili_snap::json_string!(response), @r###"
+    {
+      "vectorStore": true,
+      "metrics": false,
+      "logsRoute": false,
+      "exportPuffinReports": false
+    }
+    "###);
+
+    let (response, code) = index
+        .update_settings(json!({ "embedders": {"default": {
+                "source": "userProvided",
+                "dimensions": 2}}} ))
+        .await;
+    assert_eq!(202, code, "{:?}", response);
+    index.wait_task(response.uid()).await;
+
+    let (response, code) = index.add_documents(documents.clone(), None).await;
+    assert_eq!(202, code, "{:?}", response);
+    index.wait_task(response.uid()).await;
+    index
+}
+
+static SIMPLE_SEARCH_DOCUMENTS: Lazy<Value> = Lazy::new(|| {
+    json!([
+    {
+        "title": "Shazam!",
+        "desc": "a Captain Marvel ersatz",
+        "id": "1",
+        "_vectors": {"default": [1.0, 3.0]},
+    },
+    {
+        "title": "Captain Planet",
+        "desc": "He's not part of the Marvel Cinematic Universe",
+        "id": "2",
+        "_vectors": {"default": [1.0, 2.0]},
+    },
+    {
+        "title": "Captain Marvel",
+        "desc": "a Shazam ersatz",
+        "id": "3",
+        "_vectors": {"default": [2.0, 3.0]},
+    }])
+});
+
+static SINGLE_DOCUMENT: Lazy<Value> = Lazy::new(|| {
+    json!([{
+            "title": "Shazam!",
+            "desc": "a Captain Marvel ersatz",
+            "id": "1",
+            "_vectors": {"default": [1.0, 3.0]},
+    }])
+});
+
+#[actix_rt::test]
+async fn simple_search() {
+    let server = Server::new().await;
+    let index = index_with_documents(&server, &SIMPLE_SEARCH_DOCUMENTS).await;
+
+    let (response, code) = index
+        .search_post(
+            json!({"q": "Captain", "vector": [1.0, 1.0], "hybrid": {"semanticRatio": 0.2}}),
+        )
+        .await;
+    snapshot!(code, @"200 OK");
+    snapshot!(response["hits"], @r###"[{"title":"Captain Planet","desc":"He's not part of the Marvel Cinematic Universe","id":"2","_vectors":{"default":[1.0,2.0]}},{"title":"Captain Marvel","desc":"a Shazam ersatz","id":"3","_vectors":{"default":[2.0,3.0]}},{"title":"Shazam!","desc":"a Captain Marvel ersatz","id":"1","_vectors":{"default":[1.0,3.0]}}]"###);
+
+    let (response, code) = index
+        .search_post(
+            json!({"q": "Captain", "vector": [1.0, 1.0], "hybrid": {"semanticRatio": 0.8}}),
+        )
+        .await;
+    snapshot!(code, @"200 OK");
+    snapshot!(response["hits"], @r###"[{"title":"Captain Marvel","desc":"a Shazam ersatz","id":"3","_vectors":{"default":[2.0,3.0]},"_semanticScore":0.99029034},{"title":"Captain Planet","desc":"He's not part of the Marvel Cinematic Universe","id":"2","_vectors":{"default":[1.0,2.0]},"_semanticScore":0.97434163},{"title":"Shazam!","desc":"a Captain Marvel ersatz","id":"1","_vectors":{"default":[1.0,3.0]},"_semanticScore":0.9472136}]"###);
+}
+
+#[actix_rt::test]
+async fn highlighter() {
+    let server = Server::new().await;
+    let index = index_with_documents(&server, &SIMPLE_SEARCH_DOCUMENTS).await;
+
+    let (response, code) = index
+        .search_post(json!({"q": "Captain Marvel", "vector": [1.0, 1.0],
+            "hybrid": {"semanticRatio": 0.2},
+            "attributesToHighlight": [
+                     "desc"
+                   ],
+                   "highlightPreTag": "**BEGIN**",
+                   "highlightPostTag": "**END**"
+        }))
+        .await;
+    snapshot!(code, @"200 OK");
+    snapshot!(response["hits"], @r###"[{"title":"Captain Marvel","desc":"a Shazam ersatz","id":"3","_vectors":{"default":[2.0,3.0]},"_formatted":{"title":"Captain Marvel","desc":"a Shazam ersatz","id":"3","_vectors":{"default":["2.0","3.0"]}}},{"title":"Shazam!","desc":"a Captain Marvel ersatz","id":"1","_vectors":{"default":[1.0,3.0]},"_formatted":{"title":"Shazam!","desc":"a **BEGIN**Captain**END** **BEGIN**Marvel**END** ersatz","id":"1","_vectors":{"default":["1.0","3.0"]}}},{"title":"Captain Planet","desc":"He's not part of the Marvel Cinematic Universe","id":"2","_vectors":{"default":[1.0,2.0]},"_formatted":{"title":"Captain Planet","desc":"He's not part of the **BEGIN**Marvel**END** Cinematic Universe","id":"2","_vectors":{"default":["1.0","2.0"]}}}]"###);
+
+    let (response, code) = index
+        .search_post(json!({"q": "Captain Marvel", "vector": [1.0, 1.0],
+            "hybrid": {"semanticRatio": 0.8},
+            "attributesToHighlight": [
+                     "desc"
+                   ],
+                   "highlightPreTag": "**BEGIN**",
+                   "highlightPostTag": "**END**"
+        }))
+        .await;
+    snapshot!(code, @"200 OK");
+    snapshot!(response["hits"], @r###"[{"title":"Captain Marvel","desc":"a Shazam ersatz","id":"3","_vectors":{"default":[2.0,3.0]},"_formatted":{"title":"Captain Marvel","desc":"a Shazam ersatz","id":"3","_vectors":{"default":["2.0","3.0"]}},"_semanticScore":0.99029034},{"title":"Captain Planet","desc":"He's not part of the Marvel Cinematic Universe","id":"2","_vectors":{"default":[1.0,2.0]},"_formatted":{"title":"Captain Planet","desc":"He's not part of the **BEGIN**Marvel**END** Cinematic Universe","id":"2","_vectors":{"default":["1.0","2.0"]}},"_semanticScore":0.97434163},{"title":"Shazam!","desc":"a Captain Marvel ersatz","id":"1","_vectors":{"default":[1.0,3.0]},"_formatted":{"title":"Shazam!","desc":"a **BEGIN**Captain**END** **BEGIN**Marvel**END** ersatz","id":"1","_vectors":{"default":["1.0","3.0"]}},"_semanticScore":0.9472136}]"###);
+
+    // no highlighting on full semantic
+    let (response, code) = index
+        .search_post(json!({"q": "Captain Marvel", "vector": [1.0, 1.0],
+            "hybrid": {"semanticRatio": 1.0},
+            "attributesToHighlight": [
+                     "desc"
+                   ],
+                   "highlightPreTag": "**BEGIN**",
+                   "highlightPostTag": "**END**"
+        }))
+        .await;
+    snapshot!(code, @"200 OK");
+    snapshot!(response["hits"], @r###"[{"title":"Captain Marvel","desc":"a Shazam ersatz","id":"3","_vectors":{"default":[2.0,3.0]},"_formatted":{"title":"Captain Marvel","desc":"a Shazam ersatz","id":"3","_vectors":{"default":["2.0","3.0"]}},"_semanticScore":0.99029034},{"title":"Captain Planet","desc":"He's not part of the Marvel Cinematic Universe","id":"2","_vectors":{"default":[1.0,2.0]},"_formatted":{"title":"Captain Planet","desc":"He's not part of the Marvel Cinematic Universe","id":"2","_vectors":{"default":["1.0","2.0"]}},"_semanticScore":0.97434163},{"title":"Shazam!","desc":"a Captain Marvel ersatz","id":"1","_vectors":{"default":[1.0,3.0]},"_formatted":{"title":"Shazam!","desc":"a Captain Marvel ersatz","id":"1","_vectors":{"default":["1.0","3.0"]}}}]"###);
+}
+
+#[actix_rt::test]
+async fn invalid_semantic_ratio() {
+    let server = Server::new().await;
+    let index = index_with_documents(&server, &SIMPLE_SEARCH_DOCUMENTS).await;
+
+    let (response, code) = index
+        .search_post(
+            json!({"q": "Captain", "vector": [1.0, 1.0], "hybrid": {"semanticRatio": 1.2}}),
+        )
+        .await;
+    snapshot!(code, @"400 Bad Request");
+    snapshot!(response, @r###"
+    {
+      "message": "Invalid value at `.hybrid.semanticRatio`: the value of `semanticRatio` is invalid, expected a float between `0.0` and `1.0`.",
+      "code": "invalid_search_semantic_ratio",
+      "type": "invalid_request",
+      "link": "https://docs.meilisearch.com/errors#invalid_search_semantic_ratio"
+    }
+    "###);
+
+    let (response, code) = index
+        .search_post(
+            json!({"q": "Captain", "vector": [1.0, 1.0], "hybrid": {"semanticRatio": -0.8}}),
+        )
+        .await;
+    snapshot!(code, @"400 Bad Request");
+    snapshot!(response, @r###"
+    {
+      "message": "Invalid value at `.hybrid.semanticRatio`: the value of `semanticRatio` is invalid, expected a float between `0.0` and `1.0`.",
+      "code": "invalid_search_semantic_ratio",
+      "type": "invalid_request",
+      "link": "https://docs.meilisearch.com/errors#invalid_search_semantic_ratio"
+    }
+    "###);
+
+    let (response, code) = index
+        .search_get(
+            &yaup::to_string(
+                &json!({"q": "Captain", "vector": [1.0, 1.0], "hybridSemanticRatio": 1.2}),
+            )
+            .unwrap(),
+        )
+        .await;
+    snapshot!(code, @"400 Bad Request");
+    snapshot!(response, @r###"
+    {
+      "message": "Invalid value in parameter `hybridSemanticRatio`: the value of `semanticRatio` is invalid, expected a float between `0.0` and `1.0`.",
+      "code": "invalid_search_semantic_ratio",
+      "type": "invalid_request",
+      "link": "https://docs.meilisearch.com/errors#invalid_search_semantic_ratio"
+    }
+    "###);
+
+    let (response, code) = index
+        .search_get(
+            &yaup::to_string(
+                &json!({"q": "Captain", "vector": [1.0, 1.0], "hybridSemanticRatio": -0.2}),
+            )
+            .unwrap(),
+        )
+        .await;
+    snapshot!(code, @"400 Bad Request");
+    snapshot!(response, @r###"
+    {
+      "message": "Invalid value in parameter `hybridSemanticRatio`: the value of `semanticRatio` is invalid, expected a float between `0.0` and `1.0`.",
+      "code": "invalid_search_semantic_ratio",
+      "type": "invalid_request",
+      "link": "https://docs.meilisearch.com/errors#invalid_search_semantic_ratio"
+    }
+    "###);
+}
+
+#[actix_rt::test]
+async fn single_document() {
+    let server = Server::new().await;
+    let index = index_with_documents(&server, &SINGLE_DOCUMENT).await;
+
+    let (response, code) = index
+    .search_post(
+        json!({"vector": [1.0, 3.0], "hybrid": {"semanticRatio": 1.0}, "showRankingScore": true}),
+    )
+    .await;
+
+    snapshot!(code, @"200 OK");
+    snapshot!(response["hits"][0], @r###"{"title":"Shazam!","desc":"a Captain Marvel ersatz","id":"1","_vectors":{"default":[1.0,3.0]},"_rankingScore":1.0,"_semanticScore":1.0}"###);
+}

meilisearch/tests/search/mod.rs

@@ -6,6 +6,7 @@ mod errors;
 mod facet_search;
 mod formatted;
 mod geo;
+mod hybrid;
 mod multi;
 mod pagination;
 mod restrict_searchable;
@@ -20,22 +21,27 @@ static DOCUMENTS: Lazy<Value> = Lazy::new(|| {
         {
             "title": "Shazam!",
             "id": "287947",
+            "_vectors": { "manual": [1, 2, 3]},
         },
         {
             "title": "Captain Marvel",
             "id": "299537",
+            "_vectors": { "manual": [1, 2, 54] },
         },
         {
             "title": "Escape Room",
             "id": "522681",
+            "_vectors": { "manual": [10, -23, 32] },
         },
         {
             "title": "How to Train Your Dragon: The Hidden World",
             "id": "166428",
+            "_vectors": { "manual": [-100, 231, 32] },
         },
         {
             "title": "Gläss",
             "id": "450465",
+            "_vectors": { "manual": [-100, 340, 90] },
         }
     ])
 });
@@ -57,6 +63,7 @@ static NESTED_DOCUMENTS: Lazy<Value> = Lazy::new(|| {
                 },
             ],
             "cattos": "pésti",
+            "_vectors": { "manual": [1, 2, 3]},
         },
         {
             "id": 654,
@@ -69,12 +76,14 @@ static NESTED_DOCUMENTS: Lazy<Value> = Lazy::new(|| {
                 },
             ],
             "cattos": ["simba", "pestiféré"],
+            "_vectors": { "manual": [1, 2, 54] },
         },
         {
             "id": 750,
             "father": "romain",
             "mother": "michelle",
             "cattos": ["enigma"],
+            "_vectors": { "manual": [10, 23, 32] },
         },
         {
             "id": 951,
@@ -91,6 +100,7 @@ static NESTED_DOCUMENTS: Lazy<Value> = Lazy::new(|| {
                 },
             ],
             "cattos": ["moumoute", "gomez"],
+            "_vectors": { "manual": [10, 23, 32] },
         },
     ])
 });
@@ -756,38 +766,14 @@ async fn faceting_max_values_per_facet() {
 }
 
 #[actix_rt::test]
-async fn experimental_feature_score_details() {
+async fn test_score_details() {
     let server = Server::new().await;
     let index = server.index("test");
 
     let documents = DOCUMENTS.clone();
 
-    index.add_documents(json!(documents), None).await;
-    index.wait_task(0).await;
-
-    index
-        .search(
-            json!({
-                "q": "train dragon",
-                "showRankingScoreDetails": true,
-            }),
-            |response, code| {
-                meili_snap::snapshot!(code, @"400 Bad Request");
-                meili_snap::snapshot!(meili_snap::json_string!(response), @r###"
-                {
-                  "message": "Computing score details requires enabling the `score details` experimental feature. See https://github.com/meilisearch/product/discussions/674",
-                  "code": "feature_not_enabled",
-                  "type": "invalid_request",
-                  "link": "https://docs.meilisearch.com/errors#feature_not_enabled"
-                }
-                "###);
-            },
-        )
-        .await;
-
-    let (response, code) = server.set_features(json!({"scoreDetails": true})).await;
-    meili_snap::snapshot!(code, @"200 OK");
-    meili_snap::snapshot!(response["scoreDetails"], @"true");
+    let res = index.add_documents(json!(documents), None).await;
+    index.wait_task(res.0.uid()).await;
 
     index
         .search(
@@ -802,6 +788,13 @@ async fn experimental_feature_score_details() {
                   {
                     "title": "How to Train Your Dragon: The Hidden World",
                     "id": "166428",
+                    "_vectors": {
+                      "manual": [
+                        -100,
+                        231,
+                        32
+                      ]
+                    },
                     "_rankingScoreDetails": {
                       "words": {
                         "order": 0,
@@ -823,7 +816,7 @@ async fn experimental_feature_score_details() {
                         "order": 3,
                         "attributeRankingOrderScore": 1.0,
                         "queryWordDistanceScore": 0.8095238095238095,
-                        "score": 0.9365079365079364
+                        "score": 0.9727891156462584
                       },
                       "exactness": {
                         "order": 4,
@@ -870,13 +863,100 @@ async fn experimental_feature_vector_store() {
     meili_snap::snapshot!(code, @"200 OK");
     meili_snap::snapshot!(response["vectorStore"], @"true");
 
+    let (response, code) = index
+        .update_settings(json!({"embedders": {
+            "manual": {
+                "source": "userProvided",
+                "dimensions": 3,
+            }
+        }}))
+        .await;
+
+    meili_snap::snapshot!(response, @r###"
+    {
+      "taskUid": 1,
+      "indexUid": "test",
+      "status": "enqueued",
+      "type": "settingsUpdate",
+      "enqueuedAt": "[date]"
+    }
+    "###);
+    meili_snap::snapshot!(code, @"202 Accepted");
+    let response = index.wait_task(response.uid()).await;
+
+    meili_snap::snapshot!(meili_snap::json_string!(response["status"]), @"\"succeeded\"");
+
     let (response, code) = index
         .search_post(json!({
             "vector": [1.0, 2.0, 3.0],
         }))
         .await;
+
     meili_snap::snapshot!(code, @"200 OK");
-    meili_snap::snapshot!(meili_snap::json_string!(response["hits"]), @"[]");
+    // vector search returns all documents that don't have vectors in the last bucket, like all sorts
+    meili_snap::snapshot!(meili_snap::json_string!(response["hits"]), @r###"
+    [
+      {
+        "title": "Shazam!",
+        "id": "287947",
+        "_vectors": {
+          "manual": [
+            1,
+            2,
+            3
+          ]
+        },
+        "_semanticScore": 1.0
+      },
+      {
+        "title": "Captain Marvel",
+        "id": "299537",
+        "_vectors": {
+          "manual": [
+            1,
+            2,
+            54
+          ]
+        },
+        "_semanticScore": 0.9129112
+      },
+      {
+        "title": "Gläss",
+        "id": "450465",
+        "_vectors": {
+          "manual": [
+            -100,
+            340,
+            90
+          ]
+        },
+        "_semanticScore": 0.8106413
+      },
+      {
+        "title": "How to Train Your Dragon: The Hidden World",
+        "id": "166428",
+        "_vectors": {
+          "manual": [
+            -100,
+            231,
+            32
+          ]
+        },
+        "_semanticScore": 0.74120104
+      },
+      {
+        "title": "Escape Room",
+        "id": "522681",
+        "_vectors": {
+          "manual": [
+            10,
+            -23,
+            32
+          ]
+        }
+      }
+    ]
+    "###);
 }
 
 #[cfg(feature = "default")]
@@ -1126,7 +1206,14 @@ async fn simple_search_with_strange_synonyms() {
             [
               {
                 "title": "How to Train Your Dragon: The Hidden World",
-                "id": "166428"
+                "id": "166428",
+                "_vectors": {
+                  "manual": [
+                    -100,
+                    231,
+                    32
+                  ]
+                }
               }
             ]
             "###);
@@ -1140,7 +1227,14 @@ async fn simple_search_with_strange_synonyms() {
             [
               {
                 "title": "How to Train Your Dragon: The Hidden World",
-                "id": "166428"
+                "id": "166428",
+                "_vectors": {
+                  "manual": [
+                    -100,
+                    231,
+                    32
+                  ]
+                }
               }
             ]
             "###);
@@ -1154,7 +1248,14 @@ async fn simple_search_with_strange_synonyms() {
             [
               {
                 "title": "How to Train Your Dragon: The Hidden World",
-                "id": "166428"
+                "id": "166428",
+                "_vectors": {
+                  "manual": [
+                    -100,
+                    231,
+                    32
+                  ]
+                }
               }
             ]
             "###);

meilisearch/tests/search/multi.rs

@@ -72,7 +72,14 @@ async fn simple_search_single_index() {
         "hits": [
           {
             "title": "Gläss",
-            "id": "450465"
+            "id": "450465",
+            "_vectors": {
+              "manual": [
+                -100,
+                340,
+                90
+              ]
+            }
           }
         ],
         "query": "glass",
@@ -86,7 +93,14 @@ async fn simple_search_single_index() {
         "hits": [
           {
             "title": "Captain Marvel",
-            "id": "299537"
+            "id": "299537",
+            "_vectors": {
+              "manual": [
+                1,
+                2,
+                54
+              ]
+            }
           }
         ],
         "query": "captain",
@@ -177,7 +191,14 @@ async fn simple_search_two_indexes() {
         "hits": [
           {
             "title": "Gläss",
-            "id": "450465"
+            "id": "450465",
+            "_vectors": {
+              "manual": [
+                -100,
+                340,
+                90
+              ]
+            }
           }
         ],
         "query": "glass",
@@ -203,7 +224,14 @@ async fn simple_search_two_indexes() {
                 "age": 4
               }
             ],
-            "cattos": "pésti"
+            "cattos": "pésti",
+            "_vectors": {
+              "manual": [
+                1,
+                2,
+                3
+              ]
+            }
           },
           {
             "id": 654,
@@ -218,7 +246,14 @@ async fn simple_search_two_indexes() {
             "cattos": [
               "simba",
               "pestiféré"
-            ]
+            ],
+            "_vectors": {
+              "manual": [
+                1,
+                2,
+                54
+              ]
+            }
           }
         ],
         "query": "pésti",

meilisearch/tests/search/restrict_searchable.rs

@@ -335,3 +335,35 @@ async fn exactness_ranking_rule_order() {
         })
         .await;
 }
+
+#[actix_rt::test]
+async fn search_on_exact_field() {
+    let server = Server::new().await;
+    let index = index_with_documents(
+        &server,
+        &json!([
+        {
+            "title": "Captain Marvel",
+            "exact": "Captain Marivel",
+            "id": "1",
+        },
+        {
+            "title": "Captain Marivel",
+            "exact": "Captain the Marvel",
+            "id": "2",
+        }]),
+    )
+    .await;
+
+    let (response, code) =
+        index.update_settings_typo_tolerance(json!({ "disableOnAttributes": ["exact"] })).await;
+    assert_eq!(202, code, "{:?}", response);
+    index.wait_task(1).await;
+    // Searching on an exact attribute should only return the document matching without typo.
+    index
+        .search(json!({"q": "Marvel", "attributesToSearchOn": ["exact"]}), |response, code| {
+            snapshot!(code, @"200 OK");
+            snapshot!(response["hits"].as_array().unwrap().len(), @"1");
+        })
+        .await;
+}

meilisearch/tests/settings/get_settings.rs

@@ -83,6 +83,7 @@ async fn get_settings() {
             "maxTotalHits": 1000,
         })
     );
+    assert_eq!(settings["proximityPrecision"], json!("byWord"));
 }
 
 #[actix_rt::test]

meilisearch/tests/settings/proximity_settings.rs

@@ -27,17 +27,6 @@ static DOCUMENTS: Lazy<crate::common::Value> = Lazy::new(|| {
 #[actix_rt::test]
 async fn attribute_scale_search() {
     let server = Server::new().await;
-    let (response, code) = server.set_features(json!({"proximityPrecision": true})).await;
-    meili_snap::snapshot!(code, @"200 OK");
-    meili_snap::snapshot!(meili_snap::json_string!(response), @r###"
-    {
-      "scoreDetails": false,
-      "vectorStore": false,
-      "metrics": false,
-      "exportPuffinReports": false,
-      "proximityPrecision": true
-    }
-    "###);
     let index = server.index("test");
 
     index.add_documents(DOCUMENTS.clone(), None).await;
@@ -45,7 +34,7 @@ async fn attribute_scale_search() {
 
     let (response, code) = index
         .update_settings(json!({
-            "proximityPrecision": "attributeScale",
+            "proximityPrecision": "byAttribute",
             "rankingRules": ["words", "typo", "proximity"],
         }))
         .await;
@@ -111,17 +100,6 @@ async fn attribute_scale_search() {
 #[actix_rt::test]
 async fn attribute_scale_phrase_search() {
     let server = Server::new().await;
-    let (response, code) = server.set_features(json!({"proximityPrecision": true})).await;
-    meili_snap::snapshot!(code, @"200 OK");
-    meili_snap::snapshot!(meili_snap::json_string!(response), @r###"
-    {
-      "scoreDetails": false,
-      "vectorStore": false,
-      "metrics": false,
-      "exportPuffinReports": false,
-      "proximityPrecision": true
-    }
-    "###);
     let index = server.index("test");
 
     index.add_documents(DOCUMENTS.clone(), None).await;
@@ -129,7 +107,7 @@ async fn attribute_scale_phrase_search() {
 
     let (_response, _code) = index
         .update_settings(json!({
-            "proximityPrecision": "attributeScale",
+            "proximityPrecision": "byAttribute",
             "rankingRules": ["words", "typo", "proximity"],
         }))
         .await;
@@ -190,17 +168,6 @@ async fn attribute_scale_phrase_search() {
 #[actix_rt::test]
 async fn word_scale_set_and_reset() {
     let server = Server::new().await;
-    let (response, code) = server.set_features(json!({"proximityPrecision": true})).await;
-    meili_snap::snapshot!(code, @"200 OK");
-    meili_snap::snapshot!(meili_snap::json_string!(response), @r###"
-    {
-      "scoreDetails": false,
-      "vectorStore": false,
-      "metrics": false,
-      "exportPuffinReports": false,
-      "proximityPrecision": true
-    }
-    "###);
     let index = server.index("test");
 
     index.add_documents(DOCUMENTS.clone(), None).await;
@@ -209,7 +176,7 @@ async fn word_scale_set_and_reset() {
     // Set and reset the setting ensuring the swap between the 2 settings is applied.
     let (_response, _code) = index
         .update_settings(json!({
-            "proximityPrecision": "attributeScale",
+            "proximityPrecision": "byAttribute",
             "rankingRules": ["words", "typo", "proximity"],
         }))
         .await;
@@ -217,7 +184,7 @@ async fn word_scale_set_and_reset() {
 
     let (_response, _code) = index
         .update_settings(json!({
-            "proximityPrecision": "wordScale",
+            "proximityPrecision": "byWord",
             "rankingRules": ["words", "typo", "proximity"],
         }))
         .await;
@@ -316,17 +283,6 @@ async fn word_scale_set_and_reset() {
 #[actix_rt::test]
 async fn attribute_scale_default_ranking_rules() {
     let server = Server::new().await;
-    let (response, code) = server.set_features(json!({"proximityPrecision": true})).await;
-    meili_snap::snapshot!(code, @"200 OK");
-    meili_snap::snapshot!(meili_snap::json_string!(response), @r###"
-    {
-      "scoreDetails": false,
-      "vectorStore": false,
-      "metrics": false,
-      "exportPuffinReports": false,
-      "proximityPrecision": true
-    }
-    "###);
     let index = server.index("test");
 
     index.add_documents(DOCUMENTS.clone(), None).await;
@@ -334,7 +290,7 @@ async fn attribute_scale_default_ranking_rules() {
 
     let (response, code) = index
         .update_settings(json!({
-            "proximityPrecision": "attributeScale"
+            "proximityPrecision": "byAttribute"
         }))
         .await;
     assert_eq!("202", code.as_str(), "{:?}", response);

meilisearch/tests/tasks/mod.rs

@@ -1,4 +1,5 @@
 mod errors;
+mod webhook;
 
 use meili_snap::insta::assert_json_snapshot;
 use time::format_description::well_known::Rfc3339;

meilisearch/tests/tasks/webhook.rs

@@ -0,0 +1,119 @@
+//! To test the webhook, we need to spawn a new server with a URL listening for
+//! post requests. The webhook handle starts a server and forwards all the
+//! received requests into a channel for you to handle.
+
+use std::sync::Arc;
+
+use actix_http::body::MessageBody;
+use actix_web::dev::{ServiceFactory, ServiceResponse};
+use actix_web::web::{Bytes, Data};
+use actix_web::{post, App, HttpResponse, HttpServer};
+use meili_snap::{json_string, snapshot};
+use meilisearch::Opt;
+use tokio::sync::mpsc;
+use url::Url;
+
+use crate::common::{default_settings, Server};
+use crate::json;
+
+#[post("/")]
+async fn forward_body(sender: Data<mpsc::UnboundedSender<Vec<u8>>>, body: Bytes) -> HttpResponse {
+    let body = body.to_vec();
+    sender.send(body).unwrap();
+    HttpResponse::Ok().into()
+}
+
+fn create_app(
+    sender: Arc<mpsc::UnboundedSender<Vec<u8>>>,
+) -> actix_web::App<
+    impl ServiceFactory<
+        actix_web::dev::ServiceRequest,
+        Config = (),
+        Response = ServiceResponse<impl MessageBody>,
+        Error = actix_web::Error,
+        InitError = (),
+    >,
+> {
+    App::new().service(forward_body).app_data(Data::from(sender))
+}
+
+struct WebhookHandle {
+    pub server_handle: tokio::task::JoinHandle<Result<(), std::io::Error>>,
+    pub url: String,
+    pub receiver: mpsc::UnboundedReceiver<Vec<u8>>,
+}
+
+async fn create_webhook_server() -> WebhookHandle {
+    let (sender, receiver) = mpsc::unbounded_channel();
+    let sender = Arc::new(sender);
+
+    // By listening on the port 0, the system will give us any available port.
+    let server =
+        HttpServer::new(move || create_app(sender.clone())).bind(("127.0.0.1", 0)).unwrap();
+    let (ip, scheme) = server.addrs_with_scheme()[0];
+    let url = format!("{scheme}://{ip}/");
+
+    let server_handle = tokio::spawn(server.run());
+    WebhookHandle { server_handle, url, receiver }
+}
+
+#[actix_web::test]
+async fn test_basic_webhook() {
+    let WebhookHandle { server_handle, url, mut receiver } = create_webhook_server().await;
+
+    let db_path = tempfile::tempdir().unwrap();
+    let server = Server::new_with_options(Opt {
+        task_webhook_url: Some(Url::parse(&url).unwrap()),
+        ..default_settings(db_path.path())
+    })
+    .await
+    .unwrap();
+
+    let index = server.index("tamo");
+    // May be flaky: we're relying on the fact that while the first document addition is processed, the other
+    // operations will be received and will be batched together. If it doesn't happen it's not a problem
+    // the rest of the test won't assume anything about the number of tasks per batch.
+    for i in 0..5 {
+        let (_, _status) = index.add_documents(json!({ "id": i, "doggo": "bone" }), None).await;
+    }
+
+    let mut nb_tasks = 0;
+    while let Some(payload) = receiver.recv().await {
+        let payload = String::from_utf8(payload).unwrap();
+        let jsonl = payload.split('\n');
+        for json in jsonl {
+            if json.is_empty() {
+                break; // we reached EOF
+            }
+            nb_tasks += 1;
+            let json: serde_json::Value = serde_json::from_str(json).unwrap();
+            snapshot!(
+                json_string!(json, { ".uid" => "[uid]", ".duration" => "[duration]", ".enqueuedAt" => "[date]", ".startedAt" => "[date]", ".finishedAt" => "[date]" }),
+            @r###"
+            {
+              "uid": "[uid]",
+              "indexUid": "tamo",
+              "status": "succeeded",
+              "type": "documentAdditionOrUpdate",
+              "canceledBy": null,
+              "details": {
+                "receivedDocuments": 1,
+                "indexedDocuments": 1
+              },
+              "error": null,
+              "duration": "[duration]",
+              "enqueuedAt": "[date]",
+              "startedAt": "[date]",
+              "finishedAt": "[date]"
+            }
+            "###);
+        }
+        if nb_tasks == 5 {
+            break;
+        }
+    }
+
+    assert!(nb_tasks == 5, "We should have received the 5 tasks but only received {nb_tasks}");
+
+    server_handle.abort();
+}

meilitool/Cargo.toml

@@ -9,11 +9,11 @@ edition.workspace = true
 license.workspace = true
 
 [dependencies]
-anyhow = "1.0.75"
-clap = { version = "4.2.1", features = ["derive"] }
+anyhow = "1.0.79"
+clap = { version = "4.4.17", features = ["derive"] }
 dump = { path = "../dump" }
 file-store = { path = "../file-store" }
 meilisearch-auth = { path = "../meilisearch-auth" }
 meilisearch-types = { path = "../meilisearch-types" }
-time = { version = "0.3.30", features = ["formatting"] }
-uuid = { version = "1.5.0", features = ["v4"], default-features = false }
+time = { version = "0.3.31", features = ["formatting"] }
+uuid = { version = "1.6.1", features = ["v4"], default-features = false }

milli/Cargo.toml

@@ -14,51 +14,53 @@ license.workspace = true
 [dependencies]
 bimap = { version = "0.6.3", features = ["serde"] }
 bincode = "1.3.3"
-bstr = "1.4.0"
-bytemuck = { version = "1.13.1", features = ["extern_crate_alloc"] }
-byteorder = "1.4.3"
+bstr = "1.9.0"
+bytemuck = { version = "1.14.0", features = ["extern_crate_alloc"] }
+byteorder = "1.5.0"
 charabia = { version = "0.8.5", default-features = false }
 concat-arrays = "0.1.2"
-crossbeam-channel = "0.5.8"
-deserr = "0.6.0"
-either = { version = "1.8.1", features = ["serde"] }
+crossbeam-channel = "0.5.11"
+deserr = "0.6.1"
+either = { version = "1.9.0", features = ["serde"] }
 flatten-serde-json = { path = "../flatten-serde-json" }
 fst = "0.4.7"
 fxhash = "0.2.1"
 geoutils = "0.5.1"
 grenad = { version = "0.4.5", default-features = false, features = [
-    "rayon", "tempfile"
+    "rayon",
+    "tempfile",
 ] }
 heed = { version = "0.20.0-alpha.9", default-features = false, features = [
-    "serde-json", "serde-bincode", "read-txn-no-tls"
+    "serde-json",
+    "serde-bincode",
+    "read-txn-no-tls",
 ] }
-indexmap = { version = "2.0.0", features = ["serde"] }
-instant-distance = { version = "0.6.1", features = ["with-serde"] }
+indexmap = { version = "2.1.0", features = ["serde"] }
 json-depth-checker = { path = "../json-depth-checker" }
 levenshtein_automata = { version = "0.2.1", features = ["fst_automaton"] }
 memmap2 = "0.7.1"
-obkv = "0.2.0"
-once_cell = "1.17.1"
-ordered-float = "3.6.0"
+obkv = "0.2.1"
+once_cell = "1.19.0"
+ordered-float = "4.2.0"
 rand_pcg = { version = "0.3.1", features = ["serde1"] }
-rayon = "1.7.0"
-roaring = "0.10.1"
+rayon = "1.8.0"
+roaring = "0.10.2"
 rstar = { version = "0.11.0", features = ["serde"] }
-serde = { version = "1.0.160", features = ["derive"] }
-serde_json = { version = "1.0.95", features = ["preserve_order"] }
-slice-group-by = "0.3.0"
+serde = { version = "1.0.195", features = ["derive"] }
+serde_json = { version = "1.0.111", features = ["preserve_order"] }
+slice-group-by = "0.3.1"
 smallstr = { version = "0.3.0", features = ["serde"] }
-smallvec = "1.10.0"
+smallvec = "1.12.0"
 smartstring = "1.0.1"
-tempfile = "3.5.0"
-thiserror = "1.0.40"
-time = { version = "0.3.20", features = [
+tempfile = "3.9.0"
+thiserror = "1.0.56"
+time = { version = "0.3.31", features = [
     "serde-well-known",
     "formatting",
     "parsing",
     "macros",
 ] }
-uuid = { version = "1.3.1", features = ["v4"] }
+uuid = { version = "1.6.1", features = ["v4"] }
 
 filter-parser = { path = "../filter-parser" }
 
@@ -69,14 +71,31 @@ itertools = "0.11.0"
 puffin = "0.16.0"
 
 # logging
-log = "0.4.17"
 logging_timer = "1.1.0"
-csv = "1.2.1"
+csv = "1.3.0"
+candle-core = { git = "https://github.com/huggingface/candle.git", version = "0.3.1" }
+candle-transformers = { git = "https://github.com/huggingface/candle.git", version = "0.3.1" }
+candle-nn = { git = "https://github.com/huggingface/candle.git", version = "0.3.1" }
+tokenizers = { git = "https://github.com/huggingface/tokenizers.git", tag = "v0.14.1", version = "0.14.1", default_features = false, features = ["onig"] }
+hf-hub = { git = "https://github.com/dureuill/hf-hub.git", branch = "rust_tls", default_features = false, features = [
+    "online",
+] }
+tokio = { version = "1.35.1", features = ["rt"] }
+futures = "0.3.30"
+reqwest = { version = "0.11.23", features = [
+    "rustls-tls",
+    "json",
+], default-features = false }
+tiktoken-rs = "0.5.8"
+liquid = "0.26.4"
+arroy = "0.2.0"
+rand = "0.8.5"
+tracing = "0.1.40"
 
 [dev-dependencies]
-mimalloc = { version = "0.1.37", default-features = false }
+mimalloc = { version = "0.1.39", default-features = false }
 big_s = "1.0.2"
-insta = "1.29.0"
+insta = "1.34.0"
 maplit = "1.0.2"
 md5 = "0.7.0"
 meili-snap = { path = "../meili-snap" }
@@ -110,3 +129,6 @@ greek = ["charabia/greek"]
 
 # allow khmer specialized tokenization
 khmer = ["charabia/khmer"]
+
+# allow CUDA support, see <https://github.com/meilisearch/meilisearch/issues/4306>
+cuda = ["candle-core/cuda"]

milli/examples/search.rs

@@ -5,8 +5,8 @@ use std::time::Instant;
 
 use heed::EnvOpenOptions;
 use milli::{
-    execute_search, DefaultSearchLogger, GeoSortStrategy, Index, SearchContext, SearchLogger,
-    TermsMatchingStrategy,
+    execute_search, filtered_universe, DefaultSearchLogger, GeoSortStrategy, Index, SearchContext,
+    SearchLogger, TermsMatchingStrategy,
 };
 
 #[global_allocator]
@@ -49,14 +49,15 @@ fn main() -> Result<(), Box<dyn Error>> {
             let start = Instant::now();
 
             let mut ctx = SearchContext::new(&index, &txn);
+            let universe = filtered_universe(&ctx, &None)?;
+
             let docs = execute_search(
                 &mut ctx,
-                &(!query.trim().is_empty()).then(|| query.trim().to_owned()),
-                &None,
+                (!query.trim().is_empty()).then(|| query.trim()),
                 TermsMatchingStrategy::Last,
                 milli::score_details::ScoringStrategy::Skip,
                 false,
-                &None,
+                universe,
                 &None,
                 GeoSortStrategy::default(),
                 0,

milli/src/distance.rs

@@ -1,41 +0,0 @@
-use std::ops;
-
-use instant_distance::Point;
-use serde::{Deserialize, Serialize};
-
-use crate::normalize_vector;
-
-#[derive(Debug, Default, Clone, Serialize, Deserialize)]
-pub struct NDotProductPoint(Vec<f32>);
-
-impl NDotProductPoint {
-    pub fn new(point: Vec<f32>) -> Self {
-        NDotProductPoint(normalize_vector(point))
-    }
-
-    pub fn into_inner(self) -> Vec<f32> {
-        self.0
-    }
-}
-
-impl ops::Deref for NDotProductPoint {
-    type Target = [f32];
-
-    fn deref(&self) -> &Self::Target {
-        self.0.as_slice()
-    }
-}
-
-impl Point for NDotProductPoint {
-    fn distance(&self, other: &Self) -> f32 {
-        let dist = 1.0 - dot_product_similarity(&self.0, &other.0);
-        debug_assert!(!dist.is_nan());
-        dist
-    }
-}
-
-/// Returns the dot product similarity score that will between 0.0 and 1.0
-/// if both vectors are normalized. The higher the more similar the vectors are.
-pub fn dot_product_similarity(a: &[f32], b: &[f32]) -> f32 {
-    a.iter().zip(b).map(|(a, b)| a * b).sum()
-}

milli/src/documents/reader.rs

@@ -25,6 +25,7 @@ impl<R: io::Read + io::Seek> DocumentsBatchReader<R> {
     ///
     /// It first retrieves the index, then moves to the first document. Use the `into_cursor`
     /// method to iterator over the documents, from the first to the last.
+    #[tracing::instrument(level = "trace", skip_all, target = "indexing::documents")]
     pub fn from_reader(reader: R) -> Result<Self, Error> {
         let reader = grenad::Reader::new(reader)?;
         let mut cursor = reader.into_cursor()?;

milli/src/error.rs

@@ -61,6 +61,10 @@ pub enum InternalError {
     AbortedIndexation,
     #[error("The matching words list contains at least one invalid member.")]
     InvalidMatchingWords,
+    #[error(transparent)]
+    ArroyError(#[from] arroy::Error),
+    #[error(transparent)]
+    VectorEmbeddingError(#[from] crate::vector::Error),
 }
 
 #[derive(Error, Debug)]
@@ -110,8 +114,10 @@ only composed of alphanumeric characters (a-z A-Z 0-9), hyphens (-) and undersco
     InvalidGeoField(#[from] GeoError),
     #[error("Invalid vector dimensions: expected: `{}`, found: `{}`.", .expected, .found)]
     InvalidVectorDimensions { expected: usize, found: usize },
-    #[error("The `_vectors` field in the document with the id: `{document_id}` is not an array. Was expecting an array of floats or an array of arrays of floats but instead got `{value}`.")]
-    InvalidVectorsType { document_id: Value, value: Value },
+    #[error("The `_vectors.{subfield}` field in the document with id: `{document_id}` is not an array. Was expecting an array of floats or an array of arrays of floats but instead got `{value}`.")]
+    InvalidVectorsType { document_id: Value, value: Value, subfield: String },
+    #[error("The `_vectors` field in the document with id: `{document_id}` is not an object. Was expecting an object with a key for each embedder with manually provided vectors, but instead got `{value}`")]
+    InvalidVectorsMapType { document_id: Value, value: Value },
     #[error("{0}")]
     InvalidFilter(String),
     #[error("Invalid type for filter subexpression: expected: {}, found: {1}.", .0.join(", "))]
@@ -166,7 +172,7 @@ only composed of alphanumeric characters (a-z A-Z 0-9), hyphens (-) and undersco
     TooManyDocumentIds { primary_key: String, document: Object },
     #[error("The primary key inference failed as the engine did not find any field ending with `id` in its name. Please specify the primary key manually using the `primaryKey` query parameter.")]
     NoPrimaryKeyCandidateFound,
-    #[error("The primary key inference failed as the engine found {} fields ending with `id` in their names: '{}' and '{}'. Please specify the primary key manually using the `primaryKey` query parameter.", .candidates.len(), .candidates.get(0).unwrap(), .candidates.get(1).unwrap())]
+    #[error("The primary key inference failed as the engine found {} fields ending with `id` in their names: '{}' and '{}'. Please specify the primary key manually using the `primaryKey` query parameter.", .candidates.len(), .candidates.first().unwrap(), .candidates.get(1).unwrap())]
     MultiplePrimaryKeyCandidatesFound { candidates: Vec<String> },
     #[error("There is no more space left on the device. Consider increasing the size of the disk/partition.")]
     NoSpaceLeftOnDevice,
@@ -180,6 +186,93 @@ only composed of alphanumeric characters (a-z A-Z 0-9), hyphens (-) and undersco
     UnknownInternalDocumentId { document_id: DocumentId },
     #[error("`minWordSizeForTypos` setting is invalid. `oneTypo` and `twoTypos` fields should be between `0` and `255`, and `twoTypos` should be greater or equals to `oneTypo` but found `oneTypo: {0}` and twoTypos: {1}`.")]
     InvalidMinTypoWordLenSetting(u8, u8),
+    #[error(transparent)]
+    VectorEmbeddingError(#[from] crate::vector::Error),
+    #[error(transparent)]
+    MissingDocumentField(#[from] crate::prompt::error::RenderPromptError),
+    #[error(transparent)]
+    InvalidPrompt(#[from] crate::prompt::error::NewPromptError),
+    #[error("`.embedders.{0}.documentTemplate`: Invalid template: {1}.")]
+    InvalidPromptForEmbeddings(String, crate::prompt::error::NewPromptError),
+    #[error("Too many embedders in the configuration. Found {0}, but limited to 256.")]
+    TooManyEmbedders(usize),
+    #[error("Cannot find embedder with name {0}.")]
+    InvalidEmbedder(String),
+    #[error("Too many vectors for document with id {0}: found {1}, but limited to 256.")]
+    TooManyVectors(String, usize),
+    #[error("`.embedders.{embedder_name}`: Field `{field}` unavailable for source `{source_}` (only available for sources: {}). Available fields: {}",
+        allowed_sources_for_field
+         .iter()
+         .map(|accepted| format!("`{}`", accepted))
+         .collect::<Vec<String>>()
+         .join(", "),
+        allowed_fields_for_source
+         .iter()
+         .map(|accepted| format!("`{}`", accepted))
+         .collect::<Vec<String>>()
+         .join(", ")
+    )]
+    InvalidFieldForSource {
+        embedder_name: String,
+        source_: crate::vector::settings::EmbedderSource,
+        field: &'static str,
+        allowed_fields_for_source: &'static [&'static str],
+        allowed_sources_for_field: &'static [crate::vector::settings::EmbedderSource],
+    },
+    #[error("`.embedders.{embedder_name}.model`: Invalid model `{model}` for OpenAI. Supported models: {:?}", crate::vector::openai::EmbeddingModel::supported_models())]
+    InvalidOpenAiModel { embedder_name: String, model: String },
+    #[error("`.embedders.{embedder_name}`: Missing field `{field}` (note: this field is mandatory for source {source_})")]
+    MissingFieldForSource {
+        field: &'static str,
+        source_: crate::vector::settings::EmbedderSource,
+        embedder_name: String,
+    },
+    #[error("`.embedders.{embedder_name}.dimensions`: Model `{model}` does not support overriding its native dimensions of {expected_dimensions}. Found {dimensions}")]
+    InvalidOpenAiModelDimensions {
+        embedder_name: String,
+        model: &'static str,
+        dimensions: usize,
+        expected_dimensions: usize,
+    },
+    #[error("`.embedders.{embedder_name}.dimensions`: Model `{model}` does not support overriding its dimensions to a value higher than {max_dimensions}. Found {dimensions}")]
+    InvalidOpenAiModelDimensionsMax {
+        embedder_name: String,
+        model: &'static str,
+        dimensions: usize,
+        max_dimensions: usize,
+    },
+    #[error("`.embedders.{embedder_name}.dimensions`: `dimensions` cannot be zero")]
+    InvalidSettingsDimensions { embedder_name: String },
+}
+
+impl From<crate::vector::Error> for Error {
+    fn from(value: crate::vector::Error) -> Self {
+        match value.fault() {
+            FaultSource::User => Error::UserError(value.into()),
+            FaultSource::Runtime => Error::InternalError(value.into()),
+            FaultSource::Bug => Error::InternalError(value.into()),
+            FaultSource::Undecided => Error::InternalError(value.into()),
+        }
+    }
+}
+
+impl From<arroy::Error> for Error {
+    fn from(value: arroy::Error) -> Self {
+        match value {
+            arroy::Error::Heed(heed) => heed.into(),
+            arroy::Error::Io(io) => io.into(),
+            arroy::Error::InvalidVecDimension { expected, received } => {
+                Error::UserError(UserError::InvalidVectorDimensions { expected, found: received })
+            }
+            arroy::Error::DatabaseFull
+            | arroy::Error::InvalidItemAppend
+            | arroy::Error::UnmatchingDistance { .. }
+            | arroy::Error::MissingNode
+            | arroy::Error::MissingMetadata => {
+                Error::InternalError(InternalError::ArroyError(value))
+            }
+        }
+    }
 }
 
 #[derive(Error, Debug)]
@@ -336,6 +429,26 @@ impl From<HeedError> for Error {
     }
 }
 
+#[derive(Debug, Clone, Copy)]
+pub enum FaultSource {
+    User,
+    Runtime,
+    Bug,
+    Undecided,
+}
+
+impl std::fmt::Display for FaultSource {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        let s = match self {
+            FaultSource::User => "user error",
+            FaultSource::Runtime => "runtime error",
+            FaultSource::Bug => "coding error",
+            FaultSource::Undecided => "error",
+        };
+        f.write_str(s)
+    }
+}
+
 #[test]
 fn conditionally_lookup_for_error_message() {
     let prefix = "Attribute `name` is not sortable.";

milli/src/index.rs

@@ -10,7 +10,6 @@ use roaring::RoaringBitmap;
 use rstar::RTree;
 use time::OffsetDateTime;
 
-use crate::distance::NDotProductPoint;
 use crate::documents::PrimaryKey;
 use crate::error::{InternalError, UserError};
 use crate::fields_ids_map::FieldsIdsMap;
@@ -22,7 +21,7 @@ use crate::heed_codec::{
     BEU16StrCodec, FstSetCodec, ScriptLanguageCodec, StrBEU16Codec, StrRefCodec,
 };
 use crate::proximity::ProximityPrecision;
-use crate::readable_slices::ReadableSlices;
+use crate::vector::EmbeddingConfig;
 use crate::{
     default_criteria, CboRoaringBitmapCodec, Criterion, DocumentId, ExternalDocumentsIds,
     FacetDistribution, FieldDistribution, FieldId, FieldIdWordCountCodec, GeoPoint, ObkvCodec,
@@ -30,9 +29,6 @@ use crate::{
     BEU32, BEU64,
 };
 
-/// The HNSW data-structure that we serialize, fill and search in.
-pub type Hnsw = instant_distance::Hnsw<NDotProductPoint>;
-
 pub const DEFAULT_MIN_WORD_LEN_ONE_TYPO: u8 = 5;
 pub const DEFAULT_MIN_WORD_LEN_TWO_TYPOS: u8 = 9;
 
@@ -48,10 +44,6 @@ pub mod main_key {
     pub const FIELDS_IDS_MAP_KEY: &str = "fields-ids-map";
     pub const GEO_FACETED_DOCUMENTS_IDS_KEY: &str = "geo-faceted-documents-ids";
     pub const GEO_RTREE_KEY: &str = "geo-rtree";
-    /// The prefix of the key that is used to store the, potential big, HNSW structure.
-    /// It is concatenated with a big-endian encoded number (non-human readable).
-    /// e.g. vector-hnsw0x0032.
-    pub const VECTOR_HNSW_KEY_PREFIX: &str = "vector-hnsw";
     pub const PRIMARY_KEY_KEY: &str = "primary-key";
     pub const SEARCHABLE_FIELDS_KEY: &str = "searchable-fields";
     pub const USER_DEFINED_SEARCHABLE_FIELDS_KEY: &str = "user-defined-searchable-fields";
@@ -74,6 +66,7 @@ pub mod main_key {
     pub const SORT_FACET_VALUES_BY: &str = "sort-facet-values-by";
     pub const PAGINATION_MAX_TOTAL_HITS: &str = "pagination-max-total-hits";
     pub const PROXIMITY_PRECISION: &str = "proximity-precision";
+    pub const EMBEDDING_CONFIGS: &str = "embedding_configs";
 }
 
 pub mod db_name {
@@ -99,7 +92,8 @@ pub mod db_name {
     pub const FACET_ID_STRING_FST: &str = "facet-id-string-fst";
     pub const FIELD_ID_DOCID_FACET_F64S: &str = "field-id-docid-facet-f64s";
     pub const FIELD_ID_DOCID_FACET_STRINGS: &str = "field-id-docid-facet-strings";
-    pub const VECTOR_ID_DOCID: &str = "vector-id-docids";
+    pub const VECTOR_EMBEDDER_CATEGORY_ID: &str = "vector-embedder-category-id";
+    pub const VECTOR_ARROY: &str = "vector-arroy";
     pub const DOCUMENTS: &str = "documents";
     pub const SCRIPT_LANGUAGE_DOCIDS: &str = "script_language_docids";
 }
@@ -166,8 +160,10 @@ pub struct Index {
     /// Maps the document id, the facet field id and the strings.
     pub field_id_docid_facet_strings: Database<FieldDocIdFacetStringCodec, Str>,
 
-    /// Maps a vector id to the document id that have it.
-    pub vector_id_docid: Database<BEU32, BEU32>,
+    /// Maps an embedder name to its id in the arroy store.
+    pub embedder_category_id: Database<Str, U8>,
+    /// Vector store based on arroy™.
+    pub vector_arroy: arroy::Database<arroy::distances::Angular>,
 
     /// Maps the document id to the document as an obkv store.
     pub(crate) documents: Database<BEU32, ObkvCodec>,
@@ -182,7 +178,7 @@ impl Index {
     ) -> Result<Index> {
         use db_name::*;
 
-        options.max_dbs(24);
+        options.max_dbs(25);
 
         let env = options.open(path)?;
         let mut wtxn = env.write_txn()?;
@@ -222,7 +218,11 @@ impl Index {
             env.create_database(&mut wtxn, Some(FIELD_ID_DOCID_FACET_F64S))?;
         let field_id_docid_facet_strings =
             env.create_database(&mut wtxn, Some(FIELD_ID_DOCID_FACET_STRINGS))?;
-        let vector_id_docid = env.create_database(&mut wtxn, Some(VECTOR_ID_DOCID))?;
+        // vector stuff
+        let embedder_category_id =
+            env.create_database(&mut wtxn, Some(VECTOR_EMBEDDER_CATEGORY_ID))?;
+        let vector_arroy = env.create_database(&mut wtxn, Some(VECTOR_ARROY))?;
+
         let documents = env.create_database(&mut wtxn, Some(DOCUMENTS))?;
         wtxn.commit()?;
 
@@ -252,7 +252,8 @@ impl Index {
             facet_id_is_empty_docids,
             field_id_docid_facet_f64s,
             field_id_docid_facet_strings,
-            vector_id_docid,
+            vector_arroy,
+            embedder_category_id,
             documents,
         })
     }
@@ -475,63 +476,6 @@ impl Index {
             None => Ok(RoaringBitmap::new()),
         }
     }
-
-    /* vector HNSW */
-
-    /// Writes the provided `hnsw`.
-    pub(crate) fn put_vector_hnsw(&self, wtxn: &mut RwTxn, hnsw: &Hnsw) -> heed::Result<()> {
-        // We must delete all the chunks before we write the new HNSW chunks.
-        self.delete_vector_hnsw(wtxn)?;
-
-        let chunk_size = 1024 * 1024 * (1024 + 512); // 1.5 GiB
-        let bytes = bincode::serialize(hnsw).map_err(Into::into).map_err(heed::Error::Encoding)?;
-        for (i, chunk) in bytes.chunks(chunk_size).enumerate() {
-            let i = i as u32;
-            let mut key = main_key::VECTOR_HNSW_KEY_PREFIX.as_bytes().to_vec();
-            key.extend_from_slice(&i.to_be_bytes());
-            self.main.remap_types::<Bytes, Bytes>().put(wtxn, &key, chunk)?;
-        }
-        Ok(())
-    }
-
-    /// Delete the `hnsw`.
-    pub(crate) fn delete_vector_hnsw(&self, wtxn: &mut RwTxn) -> heed::Result<bool> {
-        let mut iter = self
-            .main
-            .remap_types::<Bytes, DecodeIgnore>()
-            .prefix_iter_mut(wtxn, main_key::VECTOR_HNSW_KEY_PREFIX.as_bytes())?;
-        let mut deleted = false;
-        while iter.next().transpose()?.is_some() {
-            // We do not keep a reference to the key or the value.
-            unsafe { deleted |= iter.del_current()? };
-        }
-        Ok(deleted)
-    }
-
-    /// Returns the `hnsw`.
-    pub fn vector_hnsw(&self, rtxn: &RoTxn) -> Result<Option<Hnsw>> {
-        let mut slices = Vec::new();
-        for result in self
-            .main
-            .remap_types::<Str, Bytes>()
-            .prefix_iter(rtxn, main_key::VECTOR_HNSW_KEY_PREFIX)?
-        {
-            let (_, slice) = result?;
-            slices.push(slice);
-        }
-
-        if slices.is_empty() {
-            Ok(None)
-        } else {
-            let readable_slices: ReadableSlices<_> = slices.into_iter().collect();
-            Ok(Some(
-                bincode::deserialize_from(readable_slices)
-                    .map_err(Into::into)
-                    .map_err(heed::Error::Decoding)?,
-            ))
-        }
-    }
-
     /* field distribution */
 
     /// Writes the field distribution which associates every field name with
@@ -1528,6 +1472,41 @@ impl Index {
 
         Ok(script_language)
     }
+
+    pub(crate) fn put_embedding_configs(
+        &self,
+        wtxn: &mut RwTxn<'_>,
+        configs: Vec<(String, EmbeddingConfig)>,
+    ) -> heed::Result<()> {
+        self.main.remap_types::<Str, SerdeJson<Vec<(String, EmbeddingConfig)>>>().put(
+            wtxn,
+            main_key::EMBEDDING_CONFIGS,
+            &configs,
+        )
+    }
+
+    pub(crate) fn delete_embedding_configs(&self, wtxn: &mut RwTxn<'_>) -> heed::Result<bool> {
+        self.main.remap_key_type::<Str>().delete(wtxn, main_key::EMBEDDING_CONFIGS)
+    }
+
+    pub fn embedding_configs(
+        &self,
+        rtxn: &RoTxn<'_>,
+    ) -> Result<Vec<(String, crate::vector::EmbeddingConfig)>> {
+        Ok(self
+            .main
+            .remap_types::<Str, SerdeJson<Vec<(String, EmbeddingConfig)>>>()
+            .get(rtxn, main_key::EMBEDDING_CONFIGS)?
+            .unwrap_or_default())
+    }
+
+    pub fn default_embedding_name(&self, rtxn: &RoTxn<'_>) -> Result<String> {
+        let configs = self.embedding_configs(rtxn)?;
+        Ok(match configs.as_slice() {
+            [(ref first_name, _)] => first_name.clone(),
+            _ => "default".to_owned(),
+        })
+    }
 }
 
 #[cfg(test)]

milli/src/lib.rs

@@ -10,18 +10,18 @@ pub mod documents;
 
 mod asc_desc;
 mod criterion;
-pub mod distance;
 mod error;
 mod external_documents_ids;
 pub mod facet;
 mod fields_ids_map;
 pub mod heed_codec;
 pub mod index;
+pub mod prompt;
 pub mod proximity;
-mod readable_slices;
 pub mod score_details;
 mod search;
 pub mod update;
+pub mod vector;
 
 #[cfg(test)]
 #[macro_use]
@@ -32,13 +32,12 @@ use std::convert::{TryFrom, TryInto};
 use std::hash::BuildHasherDefault;
 
 use charabia::normalizer::{CharNormalizer, CompatibilityDecompositionNormalizer};
-pub use distance::dot_product_similarity;
 pub use filter_parser::{Condition, FilterCondition, Span, Token};
 use fxhash::{FxHasher32, FxHasher64};
 pub use grenad::CompressionType;
 pub use search::new::{
-    execute_search, DefaultSearchLogger, GeoSortStrategy, SearchContext, SearchLogger,
-    VisualSearchLogger,
+    execute_search, filtered_universe, DefaultSearchLogger, GeoSortStrategy, SearchContext,
+    SearchLogger, VisualSearchLogger,
 };
 use serde_json::Value;
 pub use {charabia as tokenizer, heed};

milli/src/prompt/context.rs

@@ -0,0 +1,97 @@
+use liquid::model::{
+    ArrayView, DisplayCow, KStringCow, ObjectRender, ObjectSource, State, Value as LiquidValue,
+};
+use liquid::{ObjectView, ValueView};
+
+use super::document::Document;
+use super::fields::Fields;
+use crate::FieldsIdsMap;
+
+#[derive(Debug, Clone)]
+pub struct Context<'a> {
+    document: &'a Document<'a>,
+    fields: Fields<'a>,
+}
+
+impl<'a> Context<'a> {
+    pub fn new(document: &'a Document<'a>, field_id_map: &'a FieldsIdsMap) -> Self {
+        Self { document, fields: Fields::new(document, field_id_map) }
+    }
+}
+
+impl<'a> ObjectView for Context<'a> {
+    fn as_value(&self) -> &dyn ValueView {
+        self
+    }
+
+    fn size(&self) -> i64 {
+        2
+    }
+
+    fn keys<'k>(&'k self) -> Box<dyn Iterator<Item = KStringCow<'k>> + 'k> {
+        Box::new(["doc", "fields"].iter().map(|s| KStringCow::from_static(s)))
+    }
+
+    fn values<'k>(&'k self) -> Box<dyn Iterator<Item = &'k dyn ValueView> + 'k> {
+        Box::new(
+            std::iter::once(self.document.as_value())
+                .chain(std::iter::once(self.fields.as_value())),
+        )
+    }
+
+    fn iter<'k>(&'k self) -> Box<dyn Iterator<Item = (KStringCow<'k>, &'k dyn ValueView)> + 'k> {
+        Box::new(self.keys().zip(self.values()))
+    }
+
+    fn contains_key(&self, index: &str) -> bool {
+        index == "doc" || index == "fields"
+    }
+
+    fn get<'s>(&'s self, index: &str) -> Option<&'s dyn ValueView> {
+        match index {
+            "doc" => Some(self.document.as_value()),
+            "fields" => Some(self.fields.as_value()),
+            _ => None,
+        }
+    }
+}
+
+impl<'a> ValueView for Context<'a> {
+    fn as_debug(&self) -> &dyn std::fmt::Debug {
+        self
+    }
+
+    fn render(&self) -> liquid::model::DisplayCow<'_> {
+        DisplayCow::Owned(Box::new(ObjectRender::new(self)))
+    }
+
+    fn source(&self) -> liquid::model::DisplayCow<'_> {
+        DisplayCow::Owned(Box::new(ObjectSource::new(self)))
+    }
+
+    fn type_name(&self) -> &'static str {
+        "object"
+    }
+
+    fn query_state(&self, state: liquid::model::State) -> bool {
+        match state {
+            State::Truthy => true,
+            State::DefaultValue | State::Empty | State::Blank => false,
+        }
+    }
+
+    fn to_kstr(&self) -> liquid::model::KStringCow<'_> {
+        let s = ObjectRender::new(self).to_string();
+        KStringCow::from_string(s)
+    }
+
+    fn to_value(&self) -> LiquidValue {
+        LiquidValue::Object(
+            self.iter().map(|(k, x)| (k.to_string().into(), x.to_value())).collect(),
+        )
+    }
+
+    fn as_object(&self) -> Option<&dyn ObjectView> {
+        Some(self)
+    }
+}

milli/src/prompt/document.rs

@@ -0,0 +1,131 @@
+use std::cell::OnceCell;
+use std::collections::BTreeMap;
+
+use liquid::model::{
+    DisplayCow, KString, KStringCow, ObjectRender, ObjectSource, State, Value as LiquidValue,
+};
+use liquid::{ObjectView, ValueView};
+
+use crate::update::del_add::{DelAdd, KvReaderDelAdd};
+use crate::FieldsIdsMap;
+
+#[derive(Debug, Clone)]
+pub struct Document<'a>(BTreeMap<&'a str, (&'a [u8], ParsedValue)>);
+
+#[derive(Debug, Clone)]
+struct ParsedValue(std::cell::OnceCell<LiquidValue>);
+
+impl ParsedValue {
+    fn empty() -> ParsedValue {
+        ParsedValue(OnceCell::new())
+    }
+
+    fn get(&self, raw: &[u8]) -> &LiquidValue {
+        self.0.get_or_init(|| {
+            let value: serde_json::Value = serde_json::from_slice(raw).unwrap();
+            liquid::model::to_value(&value).unwrap()
+        })
+    }
+}
+
+impl<'a> Document<'a> {
+    pub fn new(
+        data: obkv::KvReaderU16<'a>,
+        side: DelAdd,
+        inverted_field_map: &'a FieldsIdsMap,
+    ) -> Self {
+        let mut out_data = BTreeMap::new();
+        for (fid, raw) in data {
+            let obkv = KvReaderDelAdd::new(raw);
+            let Some(raw) = obkv.get(side) else {
+                continue;
+            };
+            let Some(name) = inverted_field_map.name(fid) else {
+                continue;
+            };
+            out_data.insert(name, (raw, ParsedValue::empty()));
+        }
+        Self(out_data)
+    }
+
+    fn is_empty(&self) -> bool {
+        self.0.is_empty()
+    }
+
+    fn len(&self) -> usize {
+        self.0.len()
+    }
+
+    fn iter(&self) -> impl Iterator<Item = (KString, LiquidValue)> + '_ {
+        self.0.iter().map(|(&k, (raw, data))| (k.to_owned().into(), data.get(raw).to_owned()))
+    }
+}
+
+impl<'a> ObjectView for Document<'a> {
+    fn as_value(&self) -> &dyn ValueView {
+        self
+    }
+
+    fn size(&self) -> i64 {
+        self.len() as i64
+    }
+
+    fn keys<'k>(&'k self) -> Box<dyn Iterator<Item = KStringCow<'k>> + 'k> {
+        let keys = BTreeMap::keys(&self.0).map(|&s| s.into());
+        Box::new(keys)
+    }
+
+    fn values<'k>(&'k self) -> Box<dyn Iterator<Item = &'k dyn ValueView> + 'k> {
+        Box::new(self.0.values().map(|(raw, v)| v.get(raw) as &dyn ValueView))
+    }
+
+    fn iter<'k>(&'k self) -> Box<dyn Iterator<Item = (KStringCow<'k>, &'k dyn ValueView)> + 'k> {
+        Box::new(self.0.iter().map(|(&k, (raw, data))| (k.into(), data.get(raw) as &dyn ValueView)))
+    }
+
+    fn contains_key(&self, index: &str) -> bool {
+        self.0.contains_key(index)
+    }
+
+    fn get<'s>(&'s self, index: &str) -> Option<&'s dyn ValueView> {
+        self.0.get(index).map(|(raw, v)| v.get(raw) as &dyn ValueView)
+    }
+}
+
+impl<'a> ValueView for Document<'a> {
+    fn as_debug(&self) -> &dyn std::fmt::Debug {
+        self
+    }
+
+    fn render(&self) -> liquid::model::DisplayCow<'_> {
+        DisplayCow::Owned(Box::new(ObjectRender::new(self)))
+    }
+
+    fn source(&self) -> liquid::model::DisplayCow<'_> {
+        DisplayCow::Owned(Box::new(ObjectSource::new(self)))
+    }
+
+    fn type_name(&self) -> &'static str {
+        "object"
+    }
+
+    fn query_state(&self, state: liquid::model::State) -> bool {
+        match state {
+            State::Truthy => true,
+            State::DefaultValue | State::Empty | State::Blank => self.is_empty(),
+        }
+    }
+
+    fn to_kstr(&self) -> liquid::model::KStringCow<'_> {
+        let s = ObjectRender::new(self).to_string();
+        KStringCow::from_string(s)
+    }
+
+    fn to_value(&self) -> LiquidValue {
+        LiquidValue::Object(self.iter().collect())
+    }
+
+    fn as_object(&self) -> Option<&dyn ObjectView> {
+        Some(self)
+    }
+}

milli/src/prompt/error.rs

@@ -0,0 +1,56 @@
+use crate::error::FaultSource;
+
+#[derive(Debug, thiserror::Error)]
+#[error("{fault}: {kind}")]
+pub struct NewPromptError {
+    pub kind: NewPromptErrorKind,
+    pub fault: FaultSource,
+}
+
+impl From<NewPromptError> for crate::Error {
+    fn from(value: NewPromptError) -> Self {
+        crate::Error::UserError(crate::UserError::InvalidPrompt(value))
+    }
+}
+
+impl NewPromptError {
+    pub(crate) fn cannot_parse_template(inner: liquid::Error) -> NewPromptError {
+        Self { kind: NewPromptErrorKind::CannotParseTemplate(inner), fault: FaultSource::User }
+    }
+
+    pub(crate) fn invalid_fields_in_template(inner: liquid::Error) -> NewPromptError {
+        Self { kind: NewPromptErrorKind::InvalidFieldsInTemplate(inner), fault: FaultSource::User }
+    }
+}
+
+#[derive(Debug, thiserror::Error)]
+pub enum NewPromptErrorKind {
+    #[error("cannot parse template: {0}")]
+    CannotParseTemplate(liquid::Error),
+    #[error("template contains invalid fields: {0}. Only `doc.*`, `fields[i].name`, `fields[i].value` are supported")]
+    InvalidFieldsInTemplate(liquid::Error),
+}
+
+#[derive(Debug, thiserror::Error)]
+#[error("{fault}: {kind}")]
+pub struct RenderPromptError {
+    pub kind: RenderPromptErrorKind,
+    pub fault: FaultSource,
+}
+impl RenderPromptError {
+    pub(crate) fn missing_context(inner: liquid::Error) -> RenderPromptError {
+        Self { kind: RenderPromptErrorKind::MissingContext(inner), fault: FaultSource::User }
+    }
+}
+
+#[derive(Debug, thiserror::Error)]
+pub enum RenderPromptErrorKind {
+    #[error("missing field in document: {0}")]
+    MissingContext(liquid::Error),
+}
+
+impl From<RenderPromptError> for crate::Error {
+    fn from(value: RenderPromptError) -> Self {
+        crate::Error::UserError(crate::UserError::MissingDocumentField(value))
+    }
+}

milli/src/prompt/fields.rs

@@ -0,0 +1,172 @@
+use liquid::model::{
+    ArrayView, DisplayCow, KStringCow, ObjectRender, ObjectSource, State, Value as LiquidValue,
+};
+use liquid::{ObjectView, ValueView};
+
+use super::document::Document;
+use crate::FieldsIdsMap;
+#[derive(Debug, Clone)]
+pub struct Fields<'a>(Vec<FieldValue<'a>>);
+
+impl<'a> Fields<'a> {
+    pub fn new(document: &'a Document<'a>, field_id_map: &'a FieldsIdsMap) -> Self {
+        Self(
+            std::iter::repeat(document)
+                .zip(field_id_map.iter())
+                .map(|(document, (_fid, name))| FieldValue { document, name })
+                .collect(),
+        )
+    }
+}
+
+#[derive(Debug, Clone, Copy)]
+pub struct FieldValue<'a> {
+    name: &'a str,
+    document: &'a Document<'a>,
+}
+
+impl<'a> ValueView for FieldValue<'a> {
+    fn as_debug(&self) -> &dyn std::fmt::Debug {
+        self
+    }
+
+    fn render(&self) -> liquid::model::DisplayCow<'_> {
+        DisplayCow::Owned(Box::new(ObjectRender::new(self)))
+    }
+
+    fn source(&self) -> liquid::model::DisplayCow<'_> {
+        DisplayCow::Owned(Box::new(ObjectSource::new(self)))
+    }
+
+    fn type_name(&self) -> &'static str {
+        "object"
+    }
+
+    fn query_state(&self, state: liquid::model::State) -> bool {
+        match state {
+            State::Truthy => true,
+            State::DefaultValue | State::Empty | State::Blank => self.is_empty(),
+        }
+    }
+
+    fn to_kstr(&self) -> liquid::model::KStringCow<'_> {
+        let s = ObjectRender::new(self).to_string();
+        KStringCow::from_string(s)
+    }
+
+    fn to_value(&self) -> LiquidValue {
+        LiquidValue::Object(
+            self.iter().map(|(k, v)| (k.to_string().into(), v.to_value())).collect(),
+        )
+    }
+
+    fn as_object(&self) -> Option<&dyn ObjectView> {
+        Some(self)
+    }
+}
+
+impl<'a> FieldValue<'a> {
+    pub fn name(&self) -> &&'a str {
+        &self.name
+    }
+
+    pub fn value(&self) -> &dyn ValueView {
+        self.document.get(self.name).unwrap_or(&LiquidValue::Nil)
+    }
+
+    pub fn is_empty(&self) -> bool {
+        self.size() == 0
+    }
+}
+
+impl<'a> ObjectView for FieldValue<'a> {
+    fn as_value(&self) -> &dyn ValueView {
+        self
+    }
+
+    fn size(&self) -> i64 {
+        2
+    }
+
+    fn keys<'k>(&'k self) -> Box<dyn Iterator<Item = KStringCow<'k>> + 'k> {
+        Box::new(["name", "value"].iter().map(|&x| KStringCow::from_static(x)))
+    }
+
+    fn values<'k>(&'k self) -> Box<dyn Iterator<Item = &'k dyn ValueView> + 'k> {
+        Box::new(
+            std::iter::once(self.name() as &dyn ValueView).chain(std::iter::once(self.value())),
+        )
+    }
+
+    fn iter<'k>(&'k self) -> Box<dyn Iterator<Item = (KStringCow<'k>, &'k dyn ValueView)> + 'k> {
+        Box::new(self.keys().zip(self.values()))
+    }
+
+    fn contains_key(&self, index: &str) -> bool {
+        index == "name" || index == "value"
+    }
+
+    fn get<'s>(&'s self, index: &str) -> Option<&'s dyn ValueView> {
+        match index {
+            "name" => Some(self.name()),
+            "value" => Some(self.value()),
+            _ => None,
+        }
+    }
+}
+
+impl<'a> ArrayView for Fields<'a> {
+    fn as_value(&self) -> &dyn ValueView {
+        self.0.as_value()
+    }
+
+    fn size(&self) -> i64 {
+        self.0.len() as i64
+    }
+
+    fn values<'k>(&'k self) -> Box<dyn Iterator<Item = &'k dyn ValueView> + 'k> {
+        self.0.values()
+    }
+
+    fn contains_key(&self, index: i64) -> bool {
+        self.0.contains_key(index)
+    }
+
+    fn get(&self, index: i64) -> Option<&dyn ValueView> {
+        ArrayView::get(&self.0, index)
+    }
+}
+
+impl<'a> ValueView for Fields<'a> {
+    fn as_debug(&self) -> &dyn std::fmt::Debug {
+        self
+    }
+
+    fn render(&self) -> liquid::model::DisplayCow<'_> {
+        self.0.render()
+    }
+
+    fn source(&self) -> liquid::model::DisplayCow<'_> {
+        self.0.source()
+    }
+
+    fn type_name(&self) -> &'static str {
+        self.0.type_name()
+    }
+
+    fn query_state(&self, state: liquid::model::State) -> bool {
+        self.0.query_state(state)
+    }
+
+    fn to_kstr(&self) -> liquid::model::KStringCow<'_> {
+        self.0.to_kstr()
+    }
+
+    fn to_value(&self) -> LiquidValue {
+        self.0.to_value()
+    }
+
+    fn as_array(&self) -> Option<&dyn ArrayView> {
+        Some(self)
+    }
+}

milli/src/prompt/mod.rs

@@ -0,0 +1,176 @@
+mod context;
+mod document;
+pub(crate) mod error;
+mod fields;
+mod template_checker;
+
+use std::convert::TryFrom;
+
+use error::{NewPromptError, RenderPromptError};
+
+use self::context::Context;
+use self::document::Document;
+use crate::update::del_add::DelAdd;
+use crate::FieldsIdsMap;
+
+pub struct Prompt {
+    template: liquid::Template,
+    template_text: String,
+}
+
+#[derive(Debug, Clone, serde::Serialize, serde::Deserialize)]
+pub struct PromptData {
+    pub template: String,
+}
+
+impl From<Prompt> for PromptData {
+    fn from(value: Prompt) -> Self {
+        Self { template: value.template_text }
+    }
+}
+
+impl TryFrom<PromptData> for Prompt {
+    type Error = NewPromptError;
+
+    fn try_from(value: PromptData) -> Result<Self, Self::Error> {
+        Prompt::new(value.template)
+    }
+}
+
+impl Clone for Prompt {
+    fn clone(&self) -> Self {
+        let template_text = self.template_text.clone();
+        Self { template: new_template(&template_text).unwrap(), template_text }
+    }
+}
+
+fn new_template(text: &str) -> Result<liquid::Template, liquid::Error> {
+    liquid::ParserBuilder::with_stdlib().build().unwrap().parse(text)
+}
+
+fn default_template() -> liquid::Template {
+    new_template(default_template_text()).unwrap()
+}
+
+fn default_template_text() -> &'static str {
+    "{% for field in fields %} \
+    {{ field.name }}: {{ field.value }}\n\
+    {% endfor %}"
+}
+
+impl Default for Prompt {
+    fn default() -> Self {
+        Self { template: default_template(), template_text: default_template_text().into() }
+    }
+}
+
+impl Default for PromptData {
+    fn default() -> Self {
+        Self { template: default_template_text().into() }
+    }
+}
+
+impl Prompt {
+    pub fn new(template: String) -> Result<Self, NewPromptError> {
+        let this = Self {
+            template: liquid::ParserBuilder::with_stdlib()
+                .build()
+                .unwrap()
+                .parse(&template)
+                .map_err(NewPromptError::cannot_parse_template)?,
+            template_text: template,
+        };
+
+        // render template with special object that's OK with `doc.*` and `fields.*`
+        this.template
+            .render(&template_checker::TemplateChecker)
+            .map_err(NewPromptError::invalid_fields_in_template)?;
+
+        Ok(this)
+    }
+
+    pub fn render(
+        &self,
+        document: obkv::KvReaderU16<'_>,
+        side: DelAdd,
+        field_id_map: &FieldsIdsMap,
+    ) -> Result<String, RenderPromptError> {
+        let document = Document::new(document, side, field_id_map);
+        let context = Context::new(&document, field_id_map);
+
+        self.template.render(&context).map_err(RenderPromptError::missing_context)
+    }
+}
+
+#[cfg(test)]
+mod test {
+    use super::Prompt;
+    use crate::error::FaultSource;
+    use crate::prompt::error::{NewPromptError, NewPromptErrorKind};
+
+    #[test]
+    fn default_template() {
+        // does not panic
+        Prompt::default();
+    }
+
+    #[test]
+    fn empty_template() {
+        Prompt::new("".into()).unwrap();
+    }
+
+    #[test]
+    fn template_ok() {
+        Prompt::new("{{doc.title}}: {{doc.overview}}".into()).unwrap();
+    }
+
+    #[test]
+    fn template_syntax() {
+        assert!(matches!(
+            Prompt::new("{{doc.title: {{doc.overview}}".into()),
+            Err(NewPromptError {
+                kind: NewPromptErrorKind::CannotParseTemplate(_),
+                fault: FaultSource::User
+            })
+        ));
+    }
+
+    #[test]
+    fn template_missing_doc() {
+        assert!(matches!(
+            Prompt::new("{{title}}: {{overview}}".into()),
+            Err(NewPromptError {
+                kind: NewPromptErrorKind::InvalidFieldsInTemplate(_),
+                fault: FaultSource::User
+            })
+        ));
+    }
+
+    #[test]
+    fn template_nested_doc() {
+        Prompt::new("{{doc.actor.firstName}}: {{doc.actor.lastName}}".into()).unwrap();
+    }
+
+    #[test]
+    fn template_fields() {
+        Prompt::new("{% for field in fields %}{{field}}{% endfor %}".into()).unwrap();
+    }
+
+    #[test]
+    fn template_fields_ok() {
+        Prompt::new("{% for field in fields %}{{field.name}}: {{field.value}}{% endfor %}".into())
+            .unwrap();
+    }
+
+    #[test]
+    fn template_fields_invalid() {
+        assert!(matches!(
+            // intentionally garbled field
+            Prompt::new("{% for field in fields %}{{field.vaelu}} {% endfor %}".into()),
+            Err(NewPromptError {
+                kind: NewPromptErrorKind::InvalidFieldsInTemplate(_),
+                fault: FaultSource::User
+            })
+        ));
+    }
+}

milli/src/prompt/template_checker.rs

@@ -0,0 +1,301 @@
+use liquid::model::{
+    ArrayView, DisplayCow, KStringCow, ObjectRender, ObjectSource, State, Value as LiquidValue,
+};
+use liquid::{Object, ObjectView, ValueView};
+
+#[derive(Debug)]
+pub struct TemplateChecker;
+
+#[derive(Debug)]
+pub struct DummyDoc;
+
+#[derive(Debug)]
+pub struct DummyFields;
+
+#[derive(Debug)]
+pub struct DummyField;
+
+const DUMMY_VALUE: &LiquidValue = &LiquidValue::Nil;
+
+impl ObjectView for DummyField {
+    fn as_value(&self) -> &dyn ValueView {
+        self
+    }
+
+    fn size(&self) -> i64 {
+        2
+    }
+
+    fn keys<'k>(&'k self) -> Box<dyn Iterator<Item = KStringCow<'k>> + 'k> {
+        Box::new(["name", "value"].iter().map(|s| KStringCow::from_static(s)))
+    }
+
+    fn values<'k>(&'k self) -> Box<dyn Iterator<Item = &'k dyn ValueView> + 'k> {
+        Box::new(vec![DUMMY_VALUE.as_view(), DUMMY_VALUE.as_view()].into_iter())
+    }
+
+    fn iter<'k>(&'k self) -> Box<dyn Iterator<Item = (KStringCow<'k>, &'k dyn ValueView)> + 'k> {
+        Box::new(self.keys().zip(self.values()))
+    }
+
+    fn contains_key(&self, index: &str) -> bool {
+        index == "name" || index == "value"
+    }
+
+    fn get<'s>(&'s self, index: &str) -> Option<&'s dyn ValueView> {
+        if self.contains_key(index) {
+            Some(DUMMY_VALUE.as_view())
+        } else {
+            None
+        }
+    }
+}
+
+impl ValueView for DummyField {
+    fn as_debug(&self) -> &dyn std::fmt::Debug {
+        self
+    }
+
+    fn render(&self) -> DisplayCow<'_> {
+        DUMMY_VALUE.render()
+    }
+
+    fn source(&self) -> DisplayCow<'_> {
+        DUMMY_VALUE.source()
+    }
+
+    fn type_name(&self) -> &'static str {
+        "object"
+    }
+
+    fn query_state(&self, state: State) -> bool {
+        match state {
+            State::Truthy => true,
+            State::DefaultValue => false,
+            State::Empty => false,
+            State::Blank => false,
+        }
+    }
+
+    fn to_kstr(&self) -> KStringCow<'_> {
+        DUMMY_VALUE.to_kstr()
+    }
+
+    fn to_value(&self) -> LiquidValue {
+        let mut this = Object::new();
+        this.insert("name".into(), LiquidValue::Nil);
+        this.insert("value".into(), LiquidValue::Nil);
+        LiquidValue::Object(this)
+    }
+
+    fn as_object(&self) -> Option<&dyn ObjectView> {
+        Some(self)
+    }
+}
+
+impl ValueView for DummyFields {
+    fn as_debug(&self) -> &dyn std::fmt::Debug {
+        self
+    }
+
+    fn render(&self) -> DisplayCow<'_> {
+        DUMMY_VALUE.render()
+    }
+
+    fn source(&self) -> DisplayCow<'_> {
+        DUMMY_VALUE.source()
+    }
+
+    fn type_name(&self) -> &'static str {
+        "array"
+    }
+
+    fn query_state(&self, state: State) -> bool {
+        match state {
+            State::Truthy => true,
+            State::DefaultValue => false,
+            State::Empty => false,
+            State::Blank => false,
+        }
+    }
+
+    fn to_kstr(&self) -> KStringCow<'_> {
+        DUMMY_VALUE.to_kstr()
+    }
+
+    fn to_value(&self) -> LiquidValue {
+        LiquidValue::Array(vec![DummyField.to_value()])
+    }
+
+    fn as_array(&self) -> Option<&dyn ArrayView> {
+        Some(self)
+    }
+}
+
+impl ArrayView for DummyFields {
+    fn as_value(&self) -> &dyn ValueView {
+        self
+    }
+
+    fn size(&self) -> i64 {
+        u16::MAX as i64
+    }
+
+    fn values<'k>(&'k self) -> Box<dyn Iterator<Item = &'k dyn ValueView> + 'k> {
+        Box::new(std::iter::once(DummyField.as_value()))
+    }
+
+    fn contains_key(&self, index: i64) -> bool {
+        index < self.size()
+    }
+
+    fn get(&self, _index: i64) -> Option<&dyn ValueView> {
+        Some(DummyField.as_value())
+    }
+}
+
+impl ObjectView for DummyDoc {
+    fn as_value(&self) -> &dyn ValueView {
+        self
+    }
+
+    fn size(&self) -> i64 {
+        1000
+    }
+
+    fn keys<'k>(&'k self) -> Box<dyn Iterator<Item = KStringCow<'k>> + 'k> {
+        Box::new(std::iter::empty())
+    }
+
+    fn values<'k>(&'k self) -> Box<dyn Iterator<Item = &'k dyn ValueView> + 'k> {
+        Box::new(std::iter::empty())
+    }
+
+    fn iter<'k>(&'k self) -> Box<dyn Iterator<Item = (KStringCow<'k>, &'k dyn ValueView)> + 'k> {
+        Box::new(std::iter::empty())
+    }
+
+    fn contains_key(&self, _index: &str) -> bool {
+        true
+    }
+
+    fn get<'s>(&'s self, _index: &str) -> Option<&'s dyn ValueView> {
+        // Recursively sends itself
+        Some(self)
+    }
+}
+
+impl ValueView for DummyDoc {
+    fn as_debug(&self) -> &dyn std::fmt::Debug {
+        self
+    }
+
+    fn render(&self) -> DisplayCow<'_> {
+        DUMMY_VALUE.render()
+    }
+
+    fn source(&self) -> DisplayCow<'_> {
+        DUMMY_VALUE.source()
+    }
+
+    fn type_name(&self) -> &'static str {
+        "object"
+    }
+
+    fn query_state(&self, state: State) -> bool {
+        match state {
+            State::Truthy => true,
+            State::DefaultValue => false,
+            State::Empty => false,
+            State::Blank => false,
+        }
+    }
+
+    fn to_kstr(&self) -> KStringCow<'_> {
+        DUMMY_VALUE.to_kstr()
+    }
+
+    fn to_value(&self) -> LiquidValue {
+        LiquidValue::Nil
+    }
+
+    fn as_object(&self) -> Option<&dyn ObjectView> {
+        Some(self)
+    }
+}
+
+impl ObjectView for TemplateChecker {
+    fn as_value(&self) -> &dyn ValueView {
+        self
+    }
+
+    fn size(&self) -> i64 {
+        2
+    }
+
+    fn keys<'k>(&'k self) -> Box<dyn Iterator<Item = KStringCow<'k>> + 'k> {
+        Box::new(["doc", "fields"].iter().map(|s| KStringCow::from_static(s)))
+    }
+
+    fn values<'k>(&'k self) -> Box<dyn Iterator<Item = &'k dyn ValueView> + 'k> {
+        Box::new(
+            std::iter::once(DummyDoc.as_value()).chain(std::iter::once(DummyFields.as_value())),
+        )
+    }
+
+    fn iter<'k>(&'k self) -> Box<dyn Iterator<Item = (KStringCow<'k>, &'k dyn ValueView)> + 'k> {
+        Box::new(self.keys().zip(self.values()))
+    }
+
+    fn contains_key(&self, index: &str) -> bool {
+        index == "doc" || index == "fields"
+    }
+
+    fn get<'s>(&'s self, index: &str) -> Option<&'s dyn ValueView> {
+        match index {
+            "doc" => Some(DummyDoc.as_value()),
+            "fields" => Some(DummyFields.as_value()),
+            _ => None,
+        }
+    }
+}
+
+impl ValueView for TemplateChecker {
+    fn as_debug(&self) -> &dyn std::fmt::Debug {
+        self
+    }
+
+    fn render(&self) -> liquid::model::DisplayCow<'_> {
+        DisplayCow::Owned(Box::new(ObjectRender::new(self)))
+    }
+
+    fn source(&self) -> liquid::model::DisplayCow<'_> {
+        DisplayCow::Owned(Box::new(ObjectSource::new(self)))
+    }
+
+    fn type_name(&self) -> &'static str {
+        "object"
+    }
+
+    fn query_state(&self, state: liquid::model::State) -> bool {
+        match state {
+            State::Truthy => true,
+            State::DefaultValue | State::Empty | State::Blank => false,
+        }
+    }
+
+    fn to_kstr(&self) -> liquid::model::KStringCow<'_> {
+        let s = ObjectRender::new(self).to_string();
+        KStringCow::from_string(s)
+    }
+
+    fn to_value(&self) -> LiquidValue {
+        LiquidValue::Object(
+            self.iter().map(|(k, x)| (k.to_string().into(), x.to_value())).collect(),
+        )
+    }
+
+    fn as_object(&self) -> Option<&dyn ObjectView> {
+        Some(self)
+    }
+}

milli/src/proximity.rs

@@ -32,6 +32,6 @@ pub fn path_proximity(path: &[Position]) -> u32 {
 #[serde(rename_all = "camelCase")]
 pub enum ProximityPrecision {
     #[default]
-    WordScale,
-    AttributeScale,
+    ByWord,
+    ByAttribute,
 }

milli/src/readable_slices.rs

@@ -1,85 +0,0 @@
-use std::io::{self, Read};
-use std::iter::FromIterator;
-
-pub struct ReadableSlices<A> {
-    inner: Vec<A>,
-    pos: u64,
-}
-
-impl<A> FromIterator<A> for ReadableSlices<A> {
-    fn from_iter<T: IntoIterator<Item = A>>(iter: T) -> Self {
-        ReadableSlices { inner: iter.into_iter().collect(), pos: 0 }
-    }
-}
-
-impl<A: AsRef<[u8]>> Read for ReadableSlices<A> {
-    fn read(&mut self, mut buf: &mut [u8]) -> io::Result<usize> {
-        let original_buf_len = buf.len();
-
-        // We explore the list of slices to find the one where we must start reading.
-        let mut pos = self.pos;
-        let index = match self
-            .inner
-            .iter()
-            .map(|s| s.as_ref().len() as u64)
-            .position(|size| pos.checked_sub(size).map(|p| pos = p).is_none())
-        {
-            Some(index) => index,
-            None => return Ok(0),
-        };
-
-        let mut inner_pos = pos as usize;
-        for slice in &self.inner[index..] {
-            let slice = &slice.as_ref()[inner_pos..];
-
-            if buf.len() > slice.len() {
-                // We must exhaust the current slice and go to the next one there is not enough here.
-                buf[..slice.len()].copy_from_slice(slice);
-                buf = &mut buf[slice.len()..];
-                inner_pos = 0;
-            } else {
-                // There is enough in this slice to fill the remaining bytes of the buffer.
-                // Let's break just after filling it.
-                buf.copy_from_slice(&slice[..buf.len()]);
-                buf = &mut [];
-                break;
-            }
-        }
-
-        let written = original_buf_len - buf.len();
-        self.pos += written as u64;
-        Ok(written)
-    }
-}
-
-#[cfg(test)]
-mod test {
-    use std::io::Read;
-
-    use super::ReadableSlices;
-
-    #[test]
-    fn basic() {
-        let data: Vec<_> = (0..100).collect();
-        let splits: Vec<_> = data.chunks(3).collect();
-        let mut rdslices: ReadableSlices<_> = splits.into_iter().collect();
-
-        let mut output = Vec::new();
-        let length = rdslices.read_to_end(&mut output).unwrap();
-        assert_eq!(length, data.len());
-        assert_eq!(output, data);
-    }
-
-    #[test]
-    fn small_reads() {
-        let data: Vec<_> = (0..u8::MAX).collect();
-        let splits: Vec<_> = data.chunks(27).collect();
-        let mut rdslices: ReadableSlices<_> = splits.into_iter().collect();
-
-        let buffer = &mut [0; 45];
-        let length = rdslices.read(buffer).unwrap();
-        let expected: Vec<_> = (0..buffer.len() as u8).collect();
-        assert_eq!(length, buffer.len());
-        assert_eq!(buffer, &expected[..]);
-    }
-}

milli/src/score_details.rs

@@ -1,3 +1,6 @@
+use std::cmp::Ordering;
+
+use itertools::Itertools;
 use serde::Serialize;
 
 use crate::distance_between_two_points;
@@ -12,9 +15,24 @@ pub enum ScoreDetails {
     ExactAttribute(ExactAttribute),
     ExactWords(ExactWords),
     Sort(Sort),
+    Vector(Vector),
     GeoSort(GeoSort),
 }
 
+#[derive(Clone, Copy)]
+pub enum ScoreValue<'a> {
+    Score(f64),
+    Sort(&'a Sort),
+    GeoSort(&'a GeoSort),
+}
+
+enum RankOrValue<'a> {
+    Rank(Rank),
+    Sort(&'a Sort),
+    GeoSort(&'a GeoSort),
+    Score(f64),
+}
+
 impl ScoreDetails {
     pub fn local_score(&self) -> Option<f64> {
         self.rank().map(Rank::local_score)
@@ -31,11 +49,55 @@ impl ScoreDetails {
             ScoreDetails::ExactWords(details) => Some(details.rank()),
             ScoreDetails::Sort(_) => None,
             ScoreDetails::GeoSort(_) => None,
+            ScoreDetails::Vector(_) => None,
         }
     }
 
-    pub fn global_score<'a>(details: impl Iterator<Item = &'a Self>) -> f64 {
-        Rank::global_score(details.filter_map(Self::rank))
+    pub fn global_score<'a>(details: impl Iterator<Item = &'a Self> + 'a) -> f64 {
+        Self::score_values(details)
+            .find_map(|x| {
+                let ScoreValue::Score(score) = x else {
+                    return None;
+                };
+                Some(score)
+            })
+            .unwrap_or(1.0f64)
+    }
+
+    pub fn score_values<'a>(
+        details: impl Iterator<Item = &'a Self> + 'a,
+    ) -> impl Iterator<Item = ScoreValue<'a>> + 'a {
+        details
+            .map(ScoreDetails::rank_or_value)
+            .coalesce(|left, right| match (left, right) {
+                (RankOrValue::Rank(left), RankOrValue::Rank(right)) => {
+                    Ok(RankOrValue::Rank(Rank::merge(left, right)))
+                }
+                (left, right) => Err((left, right)),
+            })
+            .map(|rank_or_value| match rank_or_value {
+                RankOrValue::Rank(r) => ScoreValue::Score(r.local_score()),
+                RankOrValue::Sort(s) => ScoreValue::Sort(s),
+                RankOrValue::GeoSort(g) => ScoreValue::GeoSort(g),
+                RankOrValue::Score(s) => ScoreValue::Score(s),
+            })
+    }
+
+    fn rank_or_value(&self) -> RankOrValue<'_> {
+        match self {
+            ScoreDetails::Words(w) => RankOrValue::Rank(w.rank()),
+            ScoreDetails::Typo(t) => RankOrValue::Rank(t.rank()),
+            ScoreDetails::Proximity(p) => RankOrValue::Rank(*p),
+            ScoreDetails::Fid(f) => RankOrValue::Rank(*f),
+            ScoreDetails::Position(p) => RankOrValue::Rank(*p),
+            ScoreDetails::ExactAttribute(e) => RankOrValue::Rank(e.rank()),
+            ScoreDetails::ExactWords(e) => RankOrValue::Rank(e.rank()),
+            ScoreDetails::Sort(sort) => RankOrValue::Sort(sort),
+            ScoreDetails::GeoSort(geosort) => RankOrValue::GeoSort(geosort),
+            ScoreDetails::Vector(vector) => RankOrValue::Score(
+                vector.value_similarity.as_ref().map(|(_, s)| *s as f64).unwrap_or(0.0f64),
+            ),
+        }
     }
 
     /// Panics
@@ -181,6 +243,19 @@ impl ScoreDetails {
                     details_map.insert(sort, sort_details);
                     order += 1;
                 }
+                ScoreDetails::Vector(s) => {
+                    let vector = format!("vectorSort({:?})", s.target_vector);
+                    let value = s.value_similarity.as_ref().map(|(v, _)| v);
+                    let similarity = s.value_similarity.as_ref().map(|(_, s)| s);
+
+                    let details = serde_json::json!({
+                        "order": order,
+                        "value": value,
+                        "similarity": similarity,
+                    });
+                    details_map.insert(vector, details);
+                    order += 1;
+                }
             }
         }
         details_map
@@ -297,15 +372,21 @@ impl Rank {
     pub fn global_score(details: impl Iterator<Item = Self>) -> f64 {
         let mut rank = Rank { rank: 1, max_rank: 1 };
         for inner_rank in details {
-            rank.rank -= 1;
-
-            rank.rank *= inner_rank.max_rank;
-            rank.max_rank *= inner_rank.max_rank;
-
-            rank.rank += inner_rank.rank;
+            rank = Rank::merge(rank, inner_rank);
         }
         rank.local_score()
     }
+
+    pub fn merge(mut outer: Rank, inner: Rank) -> Rank {
+        outer.rank = outer.rank.saturating_sub(1);
+
+        outer.rank *= inner.max_rank;
+        outer.max_rank *= inner.max_rank;
+
+        outer.rank += inner.rank;
+
+        outer
+    }
 }
 
 #[derive(Debug, Clone, Copy, PartialEq, Eq, PartialOrd, Ord, Hash, Serialize)]
@@ -335,13 +416,78 @@ pub struct Sort {
     pub value: serde_json::Value,
 }
 
-#[derive(Debug, Clone, Copy, PartialEq, PartialOrd)]
+impl PartialOrd for Sort {
+    fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
+        if self.field_name != other.field_name {
+            return None;
+        }
+        if self.ascending != other.ascending {
+            return None;
+        }
+        match (&self.value, &other.value) {
+            (serde_json::Value::Null, serde_json::Value::Null) => Some(Ordering::Equal),
+            (serde_json::Value::Null, _) => Some(Ordering::Less),
+            (_, serde_json::Value::Null) => Some(Ordering::Greater),
+            // numbers are always before strings
+            (serde_json::Value::Number(_), serde_json::Value::String(_)) => Some(Ordering::Greater),
+            (serde_json::Value::String(_), serde_json::Value::Number(_)) => Some(Ordering::Less),
+            (serde_json::Value::Number(left), serde_json::Value::Number(right)) => {
+                // FIXME: unwrap permitted here?
+                let order = left.as_f64().unwrap().partial_cmp(&right.as_f64().unwrap())?;
+                // 12 < 42, and when ascending, we want to see 12 first, so the smallest.
+                // Hence, when ascending, smaller is better
+                Some(if self.ascending { order.reverse() } else { order })
+            }
+            (serde_json::Value::String(left), serde_json::Value::String(right)) => {
+                let order = left.cmp(right);
+                // Taking e.g. "a" and "z"
+                // "a" < "z", and when ascending, we want to see "a" first, so the smallest.
+                // Hence, when ascending, smaller is better
+                Some(if self.ascending { order.reverse() } else { order })
+            }
+            _ => None,
+        }
+    }
+}
+
+#[derive(Debug, Clone, Copy, PartialEq)]
 pub struct GeoSort {
     pub target_point: [f64; 2],
     pub ascending: bool,
     pub value: Option<[f64; 2]>,
 }
 
+impl PartialOrd for GeoSort {
+    fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
+        if self.target_point != other.target_point {
+            return None;
+        }
+        if self.ascending != other.ascending {
+            return None;
+        }
+        Some(match (self.distance(), other.distance()) {
+            (None, None) => Ordering::Equal,
+            (None, Some(_)) => Ordering::Less,
+            (Some(_), None) => Ordering::Greater,
+            (Some(left), Some(right)) => {
+                let order = left.partial_cmp(&right)?;
+                if self.ascending {
+                    // when ascending, the one with the smallest distance has the best score
+                    order.reverse()
+                } else {
+                    order
+                }
+            }
+        })
+    }
+}
+
+#[derive(Debug, Clone, PartialEq, PartialOrd)]
+pub struct Vector {
+    pub target_vector: Vec<f32>,
+    pub value_similarity: Option<(Vec<f32>, f32)>,
+}
+
 impl GeoSort {
     pub fn distance(&self) -> Option<f64> {
         self.value.map(|value| distance_between_two_points(&self.target_point, &value))